Apple Inc. v. Samsung Electronics Co. Ltd. et al

Filing 426

Reply Declaration of Sanjay Sood in Support of #86 Apple's Motion for a Preliminary Injunction filed by Apple Inc.. (Attachments: #1 Exhibit A, #2 Exhibit B, #3 Exhibit C, #4 Exhibit D, #5 Exhibit E, #6 Exhibit F, #7 Exhibit G, #8 Exhibit H, #9 Exhibit I)(Related document(s) #424 ) (Bartlett, Jason) (Filed on 11/28/2011) Modified text on 11/29/2011 (dhm, COURT STAFF).

SOOD EXHIBIT B

Cognition, 49 (1993) 11-36
0010-0277/93/506.00 © 1993 - Elsevier Science Publishers B.V. All rights reserved.

Reason-based choice
Eldar Shafir*' a , Itamar Simonson b , Amos Tversky b
Department of .Psychology, Princeton University, Princeton, NJ 08544, USA
Department of Psychology, Graduate School of Business, Stanford University, Stanford, CA 94305,
USA

Abstract
This paper considers the role of reasons and arguments in the making of decisions.
It is proposed that, when faced with the need to choose, decision makers often seek
and construct reasons in order to resolve the conflict and fustify their choice, to
themselves and to others. Experiments that explore and manipulate the role of
reasons are reviewed, and other decision studies are interpreted from this perspective. The role of reasons in decision making is considered as it relates to
uncertainty, conflict, context effects, and normative decision rules.
The result is that peculiar feeling of inward unrest known as indecision. Fortunately it is too
familiar to need description, for to describe it would be impossible. As long as it lasts, with the
various objects before the attention, we are said to deliberate: and when finally the original
suggestion either prevails and makes the movement take place, or gets definitively quenched by its
antagonists, we are said to decide. . . in favor of one or the other course. The reinforcing and
inhibiting ideas meanwhile are termed the reasons or motives by which the decision is brought
about.
William James (1890/1981)
My way is to divide half a sheet of paper by a line into two columns; writing over the one Pro, and
over the other Con. Then, during three or four days' consideration, I put down under the different'
heads shbjt Hints Of the different motives, that at different times occur to me for or against the
measure. When I have thus got them all together in one view, 1 endeavor to estimate the respective ·
weights. . .find at length where the balance lies. . .And, though the weight of reasons cannol· be
taken with the precision of algebraic quantities, yet,.when each is thus considered, separately and

This research was supported by US Public Health Service Grant No! 1-R29-MH46885 from the
National Institute of Mental Health, by Grant No. 89-0064 from the Air Force Office of Scientific
Research and by Grant No. SES-9109535 from the National Science Foundation. The paper was
partially prepared while the first author participated in a Summer Institute on Negotiation and Dispute
Resolution at the Center for Advanced Study in the Behavioral Sciences, and while the second author
was at the University of California, Berkeley. Funds for support of the Summer Institute were
provided by the Andrew W. Mellon Foundation. We thank Robyn Dawes for helpful comments on an
earlier draft.
Corresponding author.

12

Shafir et αϊ. / Cognition 49 (1993) 11-36
c o m p a r a t i v e l y , and the whole matter lies before me, I t h i n k I can judge better, and am less liable
to make a rash step; and in fact I have found great advantage for this kind of equation, in w h a t
may be called moral or prudential algebra.
Benjamin Franklin, 1772 (cited in Bigelow, 1887)

Introduction
The making of decisions, both big and small, is often difficult because of
uncertainty and conflict. We are usually uncertain about the exact consequences of
our actions, which may depend on the weather or the state of the economy, and
we often experience conflict about how much of one attribute (e.g., savings) to
trade off in favor of another (e.g., leisure). In order to explain how people resolve
such conflict, students of decision making have traditionally employed either
formal models or reason-based analyses. The formal modeling approach, which is
commonly used in economics, management science, and decision research,
typically associates a numerical value with each alternative, and characterizes
choice as the maximization of value. Such value-based accounts include normative
models, like expected utility theory (von Neumann & Morgenstern, 1947), as well
as descriptive models, such as prospect theory (Kahneman & Tversky, 1979). An
alternative tradition in the study of decision making, characteristic of scholarship
in history and the law, and typical of political and business discourse, employs an
informal, reason-based analysis. This approach identifies various reasons and
arguments that are purported to enter into and influence decision, and explains
choice in terms of the balance of reasons for and against the various alternatives.
Examples of reason-based analyses can be found in studies of historic presidential
decisions, such as those taken during the Cuban missile crisis (e.g., Allison,
1971), the Camp David accords (Telhami, 1990), or the Vietnam war (e.g.,
Herman, 1982; Betts & Gelb, 1979). Furthermore, reason-based analyses are
commonly used to interpret "case studies" in business and law schools. Although
the reasons invoked by researchers may not always correspond to those that
motivated the actual decision makers, it is generally agreed that an analysis in
terms of reasons may help explain decisions, especially in contexts where valuebased models can be difficult to apply.
Little contact has been made between the two traditions, which have typically
been applied to different domains. Reason-based analyses have been used
primarily to explain non-experimental data, particularly unique historic, legal and
political decisions. In contrast, value-based approaches have played a central role
in experimental studies of preference and in standard economic analyses. The two
approaches, of course, are not incompatible: reason-based accounts may often be
translated into formal models, and formal analyses can generally be paraphrased
as reason-based accounts. In the absence of a comprehensive theory of choice,

Shafir et al. I Cognition 49 (1993) 11-36

13

both formal models and reason-based analyses may contribute to the understanding of decision making.
Both approaches have obvious strengths and limitations. The formal, valuebased models have the advantage of rigor, which facilitates the derivation of
testable implications. However, value-based models are difficult to apply to
complex, real world decisions, and they often fail to capture significant aspects of
people's deliberations. An explanation of choice based on reasons, on the other
hand, is essentially qualitative in nature and typically vague. Furthermore, almost
anything can be counted as a "reason", so that every decision may be rationalized
after the fact. To overcome this difficulty, one could ask people to report their
reasons for decision. Unfortunately, the actual reasons that guide decision may or
may not correspond to those reported by the subjects. As has been amply
documented (e.g., Nisbett & Wilson, 1977), subjects are sometimes unaware of
the precise factors that determine their choices, and generate spurious explanations when asked to account for their decisions. Indeed, doubts about the validity
of introspective reports have led many students of decision making to focus
exclusively on observed choices. Although verbal reports and introspective
accounts can provide valuable information, we use "reasons" in the present
article to describe factors or motives that affect decision, whether or not they can
be articulated or recognized by the decision maker.
Despite its limitations, a reason-based conception of choice has several
attractive features. First, a focus on reasons seems closer to the way we normally
think and talk about choices. When facing a difficult choice (e.g., between
schools, or jobs) we try to come up with reasons for and against each option - we
do not normally attempt to estimate their overall values. Second, thinking of
choice as guided by reasons provides a natural way to understand the conflict that
characterizes the making of decisions. From the perspective of reason-based
choice, conflict arises when the decision maker has good reasons for and against
each option, or conflicting reasons for competing options. Unlike numerical
values, which are easy to compare, conflicting reasons may be hard to reconcile.
An analysis based on reasons can also accommodate framing effects (Tversky &
Kahneman, 1986) and elicitation effects (Tversky, Sattath, & Slovic, 1988), which
show that preferences are sensitive to the ways in which options are described
(e.g., in terms of gains or losses), and to the methods through which preferences
are elicited (e.g., pricing versus choice). These findings, which are puzzling from
the perspective of value maximization, are easier to interpret if we assume that
different frames and elicitation procedures highlight different aspects of the
options and thus bring forth different reasons to guide decision. Finally, a
conception of choice based on reasons may incorporate comparative considerations (such as relative advantages, or anticipated regret) that typically remain
outside the purview of value maximization.
In this article, we explore the logic of reason-based choice, and test some

14

Shafir et al.

I Cognition 49 (1993) 11-36

specific hypotheses concerning the role of reasons in decision making. The article
proceeds as follows. Section 1 considers the role of reasons in choice between
equally attractive options. Section 2 explores differential reliance on reasons for
and against the selection of options. Section 3 investigates the interaction between
high and low conflict and people's tendency to seek other alternatives, whereas
section 4 considers the relation between conflict and the addition of alternatives to
the choice set. Section 5 contrasts the impact of a specific reason for choice with
that of a disjunction of reasons. Section 6 explores the role that irrelevant reasons
can play in the making of decisions. Concluding remarks are presented in section
7.
1. Choice between equally attractive options
How do decision makers resolve the conflict when faced with a choice between
two equally attractive options? To investigate this question, Slovic (1975) first had
subjects equate pairs of alternatives, and later asked them to make choices
between the equally valued alternatives in each pair. One pair, for example, were
gift packages consisting of a combination of cash and coupons. For each pair, one
component of one alternative was missing, as shown below, and subjects were
asked to determine the value of the missing component that would render the two
alternatives equally attractive. (In the following example, the value volunteered
by the subject may be, say, $10).
Gift package
Cash
Coupon book worth

$32

A

Gift package В
$20
$18

A week later, subjects were asked to choose between the two equated
alternatives. They were also asked, independently, which dimension - cash or
coupons - they considered more important. Value-based theories imply that the
two alternatives - explicitly equated for value - are equally likely to be selected.
In contrast, in the choice between gift packages above, 88% of the subjects who
had equated these alternatives for value then proceeded to choose the alternative
that was higher on the dimension that the subject considered more important.
As Slovic (1975, 1990) suggests, people seem to be following a choice
mechanism that is easy to explain and justify: choosing according to the more
important dimension provides a better reason for choice than, say, random
selection, or selection of the right-hand option. Slovic (1975) replicated the above
pattern in numerous domains, including choices between college applicants, auto
tires, baseball players, and routes to work. (For additional data and a discussion
of elicitation procedures, see Tversky et al., 1988.) All the results were consistent
with the hypothesis that people do not choose between the equated alternatives at

Shafir et al. I Cognition 49 (1993) 11-36

15

random. Instead, they resolve the conflict by selecting the alternative that is
superior on the more important dimension, which seems to provide a compelling
reason for choice.

2. Reasons pro and con
Consider having to choose one of two options or, alternatively, having to reject
one of two options. Under the standard analysis of choice, the two tasks are
interchangeable. In a binary choice situation it should not matter whether people
are asked which option they prefer, or which they would reject. Because it is the
options themselves that are assumed to matter, not the way in which they are
described, if people prefer the first they will reject the second, and vice versa.
As suggested by Franklin's opening quote, our decision will depend partially
on the weights we assign to the options' pros and cons. We propose that the
positive features of options (their pros) will loom larger when choosing, whereas
the negative features of options (their cons) will be weighted more heavily when
rejecting. It is natural to select an option because of its positive features, and to
reject an option because of its negative features. To the extent that people base
their decisions on reasons for and against the options under consideration, they
are likely to focus on reasons for choosing an option when deciding which to
choose, and to focus on reasons for rejecting an option when deciding which to
reject. This hypothesis leads to a straightforward prediction: consider two
options, an enriched option, with more positive and more negative features, and
an impoverished option, with fewer positive and fewer negative features. If
positive features are weighted more heavily when choosing than when rejecting
and negative features are weighted relatively more when rejecting than when
choosing, then an enriched option could be both chosen and rejected when
compared to an impoverished option. Let Pc and Pr denote, respectively, the
percentage of subjects who choose and who reject a particular option. If choosing
and rejecting are complementary, then the sum Рс + Р, should equal 100. On the
other hand, according to the above hypothesis, Pc + Pr should be greater than 100
for the enriched option and less than 100 for the impoverished option. This
pattern was observed by Shafir (1993). Consider, for example, the following
problem which was presented to subjects in two versions that differed only in the
bracketed questions. One half of the subjects received one version, the other half
received the other. The enriched option appears last, although the order
presented to subjects was counterbalanced.
Problem 1 (n = 170):
Imagine that you serve on the jury of an only-child sole-custody case following

16

Shafir et al. I Cognition 49 (1993) 11-36

a relatively messy divorce. The facts of the case are complicated by ambiguous
economic, social, and emotional considerations, and you decide to base your
decision entirely on the following few observations. [To which parent would
you award sole custody of the child?/Which parent would you deny sole
custody of the child?
A ward
Parent A: average income
average health
average working hours
reasonable rapport with the child
relatively stable social life
Parent B:

above-average income
very close relationship with the child
extremely active social life
lots of work-related travel
minor health problems

Deny

36%

45%

64%

55%

Parent A, the impoverished option, is quite plain - with no striking positive or
negative features. There are no particularly compelling reasons to award or deny
this parent custody of the child. Parent B, the enriched option, on the other hand,
has good reasons to be awarded custody (a very close relationship with the child
and a good income), but also good reasons to be denied sole custody (health
problems and extensive absences due to travel). To the right of the options are
the percentages of subjects who chose to award and to deny custody to each of
the parents. Parent B is the majority choice both for being awarded custody of the
child and for being denied it. As predicted, Pc + Pt for parent B (64 4- 55 = 119) is
significantly greater than 100, the value expected if choosing and rejecting were
complementary (z = 2.48, p<.02). This pattern is explained by the observation
that the enriched parent (parent B) provides more compelling reasons to be
awarded as well as denied child custody.
The above pattern has been replicated in hypothetical choices between
monetary gambles, college courses, and political candidates (Shafir, 1993). For
another example, consider the following problem, presented to half the subjects
in the "prefer" and to the other half in the "cancel" version.
Problem 2 (n = 172):
Prefer:
Imagine that you are planning a week vacation in a warm spot over spring
break. You currently have two options that are reasonably priced. The travel

Shafir et al. I Cognition 49 (1993) 11-36

17

brochure gives only a limited amount of information about the two options.
Given the information available, which vacation spot would you prefer?
Cancel:
Imagine that you are planning a week vacation in a warm spot over spring
break. You currently have two options that are reasonably priced, but you can
no longer retain your reservation in both. The travel brochure gives only a
limited amount of information about the two options. Given the information
available, which reservation do you decide to cancel?
Prefer
Spot A:

average weather
average beaches
medium-quality hotel
medium-temperature water
average nightlife

Spot B: lots of sunshine
gorgeous beaches and coral reefs
ultra-modern hotel
very cold water
very strong winds
no nightlife

Cancel

33%

52%

67%

48%

The information about the two spots is typical of the kind of information we
have available when deciding where to take our next vacation. Because it is
difficult to estimate the overall value of each spot, we are likely to seek reasons
on which to base our decision. Spot A, the impoverished option, seems
unremarkable yet unobjectionable on all counts. On the other hand, there are
obvious reasons - gorgeous beaches, an abundance of sunshine, and an ultramodern hotel-for choosing spot B. Of course, there are also compelling
reasons-cold water, winds, and a lack of nightlife - why spot B should be
rejected. We suggest that the gorgeous beaches are likely to provide a more
compelling reason when we choose than when we reject, and the lack of nightlife
is likely to play a more central role when we reject than when we choose. Indeed,
spot B's share of being preferred and rejected exceeds that of spot A (Pc + P, =
67 + 48 = 115, p<.05). These results demonstrate that options are not simply
ordered according to value, with the more attractive selected and the less
attractive rejected. Instead, it appears that the relative importance of options'
strengths and weaknesses varies with the nature of the task. As a result, we are
significantly more likely to end up in spot B when we ask ourselves which we

IS

Shafir et al. / Cognition 49 (1993) 11 -36

prefer than when we contemplate which to cancel (67% vs. 52%, ζ = 2.83,
p<.001).
One of the most basic assumptions of the rational theory of choice is the
principle of procedure invariance, which requires strategically equivalent methods
of elication to yield identical preferences (see Tversky et al., 1988, for discussion).
The choose-reject discrepancy represents a predictable failure of procedure
invariance. This phenomenon is at variance with value maximization, but is easily
understood from the point of view of reason-based choice: reasons for choosing
are more compelling when we choose than when we reject, and reasons for
rejecting matter more when we reject than when we choose.

3. Choice under conflict: seeking options
The need to choose often creates conflict: we are not sure how to trade off one
attribute relative to another or, for that matter, which attributes matter to us
most. It is a commonplace that we often attempt to resolve such conflict by
seeking reasons for choosing one option over another. At times, the conflict
between available alternatives is hard to resolve, which may lead us to seek
additional options, or to maintain the status quo. Other times, the context is such
that a comparison between alternatives generates compelling reasons to choose
one option over another. Using reasons to resolve conflict has some non-obvious
implications, which are addressed below. The present section focuses on people's
decision to seek other alternatives; the next section explores some effects of
adding options to the set under consideration.
In many contexts, we need to decide whether to opt for an available option or
search for additional alternatives. Thus, a person who wishes to buy a used car
may settle for a car that is currently available or continue searching for additional
models. Seeking new alternatives usually requires additional time and effort, and
may involve the risk of losing the previously available options. Conflict plays no
role in the classical theory of choice. In this theory, each option χ has a value u(.x)
such that, for any offered set, the decision maker selects the option with the
highest value. In particular, a person is expected to search for additional
alternatives only if the expected value of searching exceeds that of the best option
currently available. A reliance on reasons, on the other hand, entails that we
should be more likely to opt for an available option when we have a convincing
reason for its selection, and that we should be more likely to search further when
a compelling reason for choice is not readily available.
To investigate this hypothesis, Tversky and Shafir (1992b) presented subjects
with pairs of options, such as bets varying in probability and payoff, or student
apartments varying in monthly rent and distance from campus, and had subjects
choose one of the two options or, instead, request an additional option, at some

Shafir et al. I Cognition 49 (1993) 11-36

19

cost. Subjects first reviewed the entire set of 12 options (gambles or apartments)
to familiarize themselves with the available alternatives. In the study of choice
between bets some subjects then received the following problem.
Conflict:
Imagine that you are offered a choice between the following two gambles:
(x) 65% chance to win $15
(y) 30% chance to win $35
You can either select one of these gambles or you can pay $1 to add one more
gamble to the choice set. The added gamble will be selected at random from
the list you reviewed.
Other subjects received a similar problem except that option у was replaced by
option x', to yield a choice between the following.
Dominance:
(x) 65% chance to win $15
(.x') 65% chance to win $14
Subjects were asked to indicate whether they wanted to add another gamble or
select between the available alternatives. They then chose their preferred gamble
from the resulting set (with or without the added option). Subjects were
instructed that the gambles they chose would be played out and that their payoff
would be proportional to the amount of money they earned minus the fee they
paid for the added gambles.
A parallel design presented choices between hypothetical student apartments.
Some subjects received the following problem.
Conflict:
Imagine that you face a choice between two apartments with the following
characteristics:
(x) $290 a month, 25 minutes from campus
(y) $350 a month, 7 minutes from campus
Both have one bedroom and a kitchenette. You can choose now between the
two apartments or you can continue to search for apartments (to be selected at

20

Shafir el al. I Cognition 49 (1993) 11-36

random from the list you reviewed). In that case, there is some risk of losing
one or both of the apartments you have found.
Other subjects received a similar problem except that option у was replaced by
option *', to yield a choice between the following.
Dominance:
(x) $290 a month, 25 minutes from campus
(x'~) $330 a month, 25 minutes from campus
Note that in both pairs of problems the choice between х and у - the conflict
condition — is non-trivial because the xs are better on one dimension and the _ys
are better on the other. In contrast, the choice between х and x' - the dominance
condition - involves no conflict because the former strictly dominates the latter.
Thus, while there is no obvious reason to choose one option over the other in the
conflict condition, there is a decisive argument for preferring one of the two
alternatives in the dominance condition.
On average, subjects requested an additional alternative 64% of the time in the
conflict condition, and only 40% of the time in the dominance condition (p <
.05). Subjects' tendency to search for additional options, in other words, was
greater when the choice among alternatives was harder to rationalize, than when
there was a compelling reason and the decision was easy.
These data are inconsistent with the principle of value maximization. According to value maximization, a subject should search for additional alternatives if
and only if the expected (subjective) value of searching exceeds that of the best
alternative currently available. Because the best alternative offered in the
dominance condition is also available in the conflict condition, value maximization
implies that the percentage of subjects who seek an additional alternative cannot
be greater in the conflict than in the dominance condition, contrary to the
observed data.
It appears that the search for additional alternatives depends not only on the
value of the best available option, as implied by value maximization, but also on
the difficulty of choosing among the options under consideration. In situations of
dominance, for example, there are clear and indisputable reasons for choosing
one option over another (e.g., "This apartment is equally distant and I save
$40!"). Having a compelling argument for choosing one of the options over the
rest reduces the temptation to look for additional alternatives. When the choice
involves conflict, on the other hand, reasons for choosing any one of the options
are less immediately available and the decision is more difficult to justify (e.g.,
"Should I save $60 a month, or reside 18 minutes closer to campus?"). In the
absence of compelling reasons for choice, there is a greater tendency to search for
other alternatives.

Shafir et αl. I Cognition 49 (1993) 11-36

21

4. Choice under conflict: adding options

An analysis in terms of reasons can help explain observed violations of the
principle of independence of irrelevant alternatives, according to which the
preference ordering between two options should not be altered by the introduction of additional alternatives. This principle follows from the standard
assumption of value maximization, and has been routinely assumed in the analysis
of consumer choice. Despite its intuitive appeal, there is a growing body of
evidence that people's preferences depend on the context of choice, defined by
the set of options under consideration. In particular, the addition and removal of
options from the offered set can influence people's preferences among options
that were available all along. Whereas in the previous section we considered
people's tendency to seek alternatives in the context of a given set of options, in
this section we illustrate phenomena that arise through the addition of options,
and interpret them in terms of reasons for choice.
A major testable implication of value maximization is that a non-preferred
option cannot become preferred when new options are added to the offered set.
In particular, a decision maker who prefers у over the option to defer the choice
should not prefer to defer the choice when both у and х are available. That the
"market share" of an option cannot be increased by enlarging the offered set is
known as the regularity condition (see Tversky & Simonson, in press). Contrary to
regularity, numerous experimental results indicate that the tendency to defer
choice can increase with the addition of alternatives. Consider, for instance, the
degree of conflict that arises when a person is presented with one attractive option
(which he or she prefers to deferring the choice), compared to two competing
alternatives. Choosing one out of two competing alternatives can be difficult: the
mere fact that an alternative is attractive may not in itself provide a compelling
reason for its selection, because the other option may be equally attractive. The
addition of an alternative may thus make the decision harder to justify, and
increase the tendency to defer the decision.
A related phenomenon was aptly described by Thomas Schelling, who tells of
an occasion in which he had decided to buy an encyclopedia for his children. At
the bookstore, he was presented with two attractive encyclopedias and, finding it
difficult to choose between the two, ended up buying neither - this, despite the
fact that had only one encyclopedia been available he would have happily bought
it. More generally, there are situations in which people prefer each of the
available alternatives over the status quo but do not have a compelling reason for
choosing among the alternatives and, as a result, defer the decision, perhaps
indefinitely.
The phenomenon described by Schelling was demonstrated by Tversky and
Shafir (1992b) in the following pair of problems, which were presented to two
groups of students (n = 124 and 121, respectively).

22

Shafir et al. I Cognition 49 (1993) 11-36
High conflict:

Suppose you are considering buying a compact disk (CD) player, and have not
yet decided what model to buy. You pass by a store that is having a 1-day
clearance sale. They offer a popular SONY player for just $99, and a
top-of-the-line AIWA player for just $169, both well below the list price. Do
you?:
(x) buy the AIWA player.
(y) buy the SONY player.
(z) wait until you learn more about the various models.

27%
27%
46%

Low conflict:
Suppose you are considering buying a CD player, and have not yet decided
what model to buy. You pass by a store that is having a 1-day clearance sale.
They offer a popular SONY player for just $99, well below the list price. Do
you?:
(y) buy the SONY player.
(z) wait until you learn more about the various models.

66%
34%

The results indicate that people are more likely to buy a CD player in the
latter, low-conflict, condition than in the former, high-conflict, situation (p < .05).
Both models - the AIWA and the SONY - seem attractive, both are well priced,
and both are on sale. The decision maker needs to determine whether she is
better off with a cheaper, popular model, or with a more expensive and
sophisticated one. This conflict is apparently not easy to resolve, and compels
many subjects to put off the purchase until they learn more about the various
options. On the other hand, when the SONY alone is available, there are
compelling arguments for its purchase: it is a popular player, it is very well priced,
and it is on sale for 1 day only. In this situation, having good reasons to choose
the offered option, a greater majority of subjects decide to opt for the CD player
rather than delay the purchase.
The addition of a competing alternative in the preceding example increased the
tendency to delay decision. Clearly, the level of conflict and its ease of resolution
depend not only on the number of options available, but on how the options
compare. Consider, for example, the following problem, in which the original
AIWA player was replaced by an inferior model (n = 62).

Shafir et al. I Cognition 49 (1993) 11-36

23

Dominance:

Suppose you are considering buying a CD player, and have not yet decided
what model to buy. You pass by a store that is having a 1-day clearance sale.
They offer a popular SONY player for just $99, well below the list price, and
an inferior AIWA player for the regular list price of $105. Do you?:
(x'J buy the AIWA player.
(y) buy the SONY player.
(z) wait until you learn more about the various models.

3%
73%
24%

In this version, contrary to the previous high-conflict version, the AIWA player
is dominated by the SONY: it is inferior in quality and costs more. Thus, the
presence of the AIWA does not detract from the reasons for buying the SONY, it
actually supplements them: the SONY is well priced, it is on sale for 1 day only,
and it is clearly better than its competitor. As a result, the SONY is chosen more
often than before the inferior AIWA was added. The ability of an asymmetrically
dominated or relatively inferior alternative, when added to a set, to increase the
attractiveness and choice probability of the dominating option is known as the
asymmetric dominance effect (Huber, Payne, & Puto, 1982). Note that in both
the high-conflict and the dominance problems subjects were presented with two
CD players and an option to delay choice. Subjects' tendency to delay, however,
is much greater when they lack clear reasons for buying either player, than when
they have compelling reasons to buy one player and not the other (p < .005).
The above patterns violate the regularity condition, which is assumed to hold
so long as the added alternatives do not provide new and relevant information. In
the above scenario, one could argue that the added options (the superior player in
one case and the inferior player in the other) conveyed information about the
consumer's chances of finding a better deal. Recall that information considerations could not explain the search experiments of the previous section because
there subjects reviewed all the potentially available options. Nevertheless, to test
this interpretation further, Tversky and Shafir (1992b) devised a similar problem,
involving real payoffs, in which the option to defer is not available. Students
(n = 80) agreed to fill out a brief questionnaire for $1.50. Following the
questionnaire, one half of the subjects were offered the opportunity to exchange
the $1.50 (the default) for one of two prizes: a metal Zebra pen (henceforth,
Zebra), or a pair of plastic Pilot pens (henceforth, Pilot). The other half of the
subjects were only offered the opportunity to exchange the $1.50 for the Zebra.
The prizes were shown to the subjects, who were also informed that each prize
regularly costs a little over $2.00. Upon indicating their preference, subjects
received their chosen option. The results were as follows. Seventy-five per cent of

24

Shafir

et

al. I Cognition 49 (1993) 11-36

the subjects chose the Zebra over the payment when the Zebra was the only
alternative, but only 47% chose the Zebra or the Pilot when both were available
(p < .05). Faced with a tempting alternative, subjects had a compelling reason to
forego the payment: the majority took advantage of the opportunity to obtain an
attractive prize of greater value. The availability of competing alternatives of
comparable value, on the other hand, did not present an immediate reason for
choosing either alternative over the other, thus increasing the tendency to retain
the default option. Similar effects in hypothetical medical decisions made by
expert physicians are documented in Redelmeier and Shafir (1993).
In the above study the addition of a competing alternative was shown to
increase the popularity of the default option. Recall that the popularity of an
option may also be enhanced by the addition of an inferior alternative. Thus, in
accord with the asymmetric dominance effect, the tendency to prefer х over у can
be increased by adding a third alternative ζ that is clearly inferior to χ but not to у
(see Fig. 1). The phenomenon of asymmetric dominance was first demonstrated,
by Huber, Payne, and Puto (1982), in choices between hypothetical options.
Wedell (1991) reports similar findings using monetary gambles. The following
example involving real choices is taken from Simonson and Tversky (1992). One
group (n = 106) was offered a choice between $6 and an elegant Cross pen. The
pen was selected by 36% of the subjects, and the remaining 64% chose the cash.
A second group (n = 115) was given a choice among three options: $6 in cash, the
same Cross pen, and a second pen that was distinctly less attractive. Only 2% of
the subjects chose the less attractive pen, but its presence increased the
percentage of subjects who chose the Cross pen from 36% to 46% (p < .10). This
pattern again violates the regularity condition discussed earlier. Similar violations
of regularity were observed in choices among other consumer goods. In one

X
Dim a
Dim a

Dimb

Dim b
Figure 1.

A schematic representation of
asymmetric dominance. The tendency to prefer х over у can be
increased by adding an alternative,
z, that is clearly inferior to х but not
to y.

Figure 2,

A schematic representation of extremeness aversion. Option у is relatively more popular in the trinary
choice, when both х and ζ are available, than in either one of the binary
comparisons, when either χ or ζ are
removed.

Shafir et al. I Cognition 49 (1993) 11-36

25

study, subjects received descriptions and pictures of microwave ovens taken from
a "Best" catalogue. One group (n = 60) was then asked to choose between an
Emerson priced at $110, and a Panasonic priced at $180. Both items were on sale,
one third off the regular price. Here, 57% chose the Emerson and 43% chose the
Panasonic. A second group (n = 60) was presented with these options along with
a $200 Panasonic at a 10% discount. Only 13% of the subjects chose the more
expensive Panasonic, but its presence increased the percentage of subjects who
chose the less expensive Panasonic from 43% to 60% (p<.05).1
Simonson and Tversky (1992) have interpreted these observations in terms of
"tradeoff contrast". They proposed that the tendency to prefer an alternative is
enhanced or hindered depending on whether the tradeoffs within the set under
consideration are favorable or unfavorable to that alternative. A second cluster of
context effects, called extremeness aversion, which refers to the finding that,
within an offered set, options with extreme values are relatively less attractive
than options with intermediate values (Simonson, 1989). For example, consider
two-dimensional options *, y, and z, such that у lies between х and z (see Fig. 2).
Considerations of value maximization imply that the middle alternative, y, should
be relatively less popular in the trinary choice than in either one of the binary
comparisons (y compared to x, or y compared to z). Extremeness aversion, on
the other hand, yields the opposite prediction because y has small advantages and
disadvantages with respect to χ and to z, whereas both χ and z have more
extreme advantages and disadvantages with respect to each other. This pattern
was observed in several experiments. For example, subjects were shown five
35 mm cameras varying in quality and price. One group (n = 106) was then given
a choice between two cameras: a Minolta X-370 priced at $170 and a Minolta
3000i priced at $240. A second group (n = 115) was given an additional option,
the Minolta 7000i priced at $470. Subjects in the first group were split evenly
between the two options, yet 57% of the subjects in the second group chose the
middle option (Minolta 3000i), with the remaining divided about equally between
the two extreme options. Thus, the introduction of an extreme option reduced the
"market share" of the other extreme option, but not of the middle option. Note
that this effect cannot be attributed to information conveyed by the offered set
because respondents had reviewed the relevant options prior to making their
choice.
We suggest that both tradeoff contrast and extremeness aversion can be
understood in terms of reasons. Suppose a decision maker faces a choice between
'These effects of context on choice can naturally be used in sales tactics. For example, WilliamsSonoma, a mail-order business located in San Francisco, used to offer a bread-baking appliance priced
at $279. They later added a second bread-baking appliance, similar to the first but somewhat larger,
and priced at $429 - more than 50% higher than the original appliance. Not surprisingly, WilliamsSonoma did not sell many units of the new item. However, the sales of the less expensive appliance
almost doubled. (To the best of our knowledge, Williams-Sonoma did not anticipate this effect.)

26

Shafir et αl. I Cognition 49 (1993) 11-36

two alternatives, χ and y, and suppose χ is of higher quality whereas у is better
priced. This produces conflict if the decision maker finds it difficult to determine
whether the quality difference outweighs the price difference. Suppose now that
the choice set also includes a third alternative, z, that is clearly inferior to _y but
not to x. The presence of z, we suggest, provides an argument for choosing у over
.v. To the extent that the initial choice between х and у is difficult, the presence of
z may help the decision maker break the tie. In the pen study, for example, the
addition of the relatively unattractive pen, whose monetary value is unclear but
whose inferiority to the elegant Cross pen is apparent, provides a reason for
choosing the Cross pen over the cash. Similarly, in the presence of options with
extreme values on the relevant dimensions, the middle option can be seen as a
compromise choice that is easier to defend than either extremes. Indeed, verbal
protocols show that the accounts generated by subjects while making these
choices involve considerations of asymmetric advantage and compromise; furthermore, asymmetric dominance is enhanced when subjects anticipate having to
justify their decisions to others (Simonson, 1989). It is noteworthy that the
arguments leading to tradeoff contrast and extremeness aversion are comparative
in nature; they are based on the positions of the options in the choice set, hence
they cannot be readily translated into the values associated with single alternatives.
Tversky and Simonson (in press) have proposed a formal model that explains
the above findings in terms of a tournament-like process in which each option is
compared against other available options in terms of their relative advantages and
disadvantages. This model can be viewed as a formal analog of the preceding
qualitative account based on reasons for choice. Which analysis - the formal or
the qualitative - proves more useful is likely to depend, among other things, on
the nature of the problem and on the purpose of the investigation.

5. Definite versus disjunctive reasons
People sometimes encounter situations of uncertainty in which they eventually
opt for the same course of action, but for very different reasons, depending on
how the uncertainty is resolved. Thus, a student who has taken an exam may
decide to take a vacation, either to reward herself in case she passes or to console
herself in case she fails. However, as illustrated below, the student may be
reluctant to commit to a vacation while the outcome of the exam is pending. The
following problem was presented by Tversky and Shafir (1992a) to 66 undergraduate students.

Shafir et al. I Cognition 49 (1993) 11-36

27

Disjunctive version:

Imagine that you have just taken a tough qualifying examination. It is the end
of the fall quarter, you feel tired and run-down, and you are not sure that you
passed the exam. In case you failed you have to take the exam again in a
couple of months - after the Christmas holidays. You now have an opportunity
to buy a very attractive 5-day Christmas vacation package in Hawaii at an
exceptionally low price. The special offer expires tomorrow, while the exam
grade will not be available until the following day. Would you?:
(a) buy the vacation package.
(b) not buy the vacation package.
(c) pay a $5 non-refundable fee in order to retain the rights to
buy the vacation package at the same exceptional price the
day after tomorrow - after you find out whether or not you
passed the exam.

32%
7%
61%

The percentage of subjects who chose each option appears on the right. Two
additional versions, called pass and fail, were presented to two different groups of
67 students each. These two versions differed only in the expression in brackets.
Pass /fail versions:
Imagine that you have just taken a tough qualifying examination. It is the end
of the fall quarter, you feel tired and run-down, and you find out that you
[passed the exam./failed the exam. You will have to take it again in a couple of
months- after the Christmas holidays.] You now have an opportunity to buy a
very attractive 5-day Christmas vacation package in Hawaii at an exceptionally
low price. The special offer expires tomorrow. Would you?:

(a) buy the vacation package.
(b) not buy the vacation package.
(c) pay a $5 non-refundable fee in order to retain
the rights to buy the vacation package at the
same exceptional price the day after tomorrow.

Pass Fail
54% 57%
16% 12%
30% 31%

The data show that more than half of the students chose the vacation package
when they knew that they passed the exam and an even larger percentage chose
the vacation when they knew that they failed. However, when they did not know

28

Shafir et al. I Cognition 49 (1993) 11-36

whether they had passed or failed, less than one third of the students chose the
vacation and 61% were willing to pay $5 to postpone the decision until the
following day, when the results of the exam would be known.2 Once the outcome
of the exam is known, the student has good - albeit different - reasons for taking
the trip: having passed the exam, the vacation is presumably seen as a reward
following a hard but successful semester; having failed the exam, the vacation
becomes a consolation and time to recuperate before a re-examination. Not
knowing the outcome of the exam, however, the student lacks a definite reason
for going to Hawaii. Notice that the outcome of the exam will be known long
before the vacation begins. Thus, the uncertainty characterizes the actual moment
of decision, not the eventual vacation.
The indeterminacy of reasons for going to Hawaii discourages many students
from buying the vacation, even when both outcomes - passing or failing the
exam -ultimately favor this course of action. Tversky and Shafir (1992a) call the
above pattern of decisions a disjunction effect. Evidently, a disjunction of
different reasons (reward in case of success or consolation in case of failure) is
often less compelling than either definite reason alone. A significant proportion of
the students above were willing to pay, in effect, for information that was
ultimately not going to affect their decision - they would choose to go to Hawaii
in either case - but that promised to leave them with a more definite reason for
making that choice. The willingness to pay for non-instrumental information is at
variance with the classical model, in which the worth of information is determined
only by its potential to influence decision.
People's preference for definite as opposed to disjunctive reasons has significant implications in cases where the option to defer decision is not available.
Consider the following series of problems presented by Tversky and Shafir
(1992a) to 98 students.
Win/lose version:
Imagine that you have just played a game of chance that gave you a 50%
chance to win $200 and a 50% chance to lose $100. The coin was tossed and
you have [won $200/lost $100]. You are now offered a second identical gamble:
50% chance to win $200 and 50% chance to lose $100. Would you?:

"An additional group of subjects (n = 123) were presented with both the fail and the pass versions,
and asked whether or not they would buy the vacation package in each case. Two thirds of the
subjects made the same choice in the two conditions, indicating that the data for the disjunctive
version cannot be explained by the hypothesis that those who like the vacation in case they pass the
exam do not like it in case they fail, and vice versa. Note that while only one third of the subjects
made different decisions depending on the outcome of the exam, more than 60% of the subjects chose
to wait when the outcome was not known.

Shafir et al. I Cognition 49 (1993) 11-36

(a) accept the second gamble.
(b) reject the second gamble.

Won
69%
31%

29

Lost
59%
41%

The students were presented with the win version of the problem above,
followed a week later by the lose version, and 10 days after that by the following
version that is a disjunction of the previous two. The problems were embedded
among other, similar problems so that the relation between the various versions
was not transparent. Subjects were instructed to treat each decision separately.
Disjunctive version:
Imagine that you have just played a game of chance that gave you a 50%
chance to win $200 and a 50% chance to lose $100. Imagine that the coin has
already been tossed, but that you will not know whether you have won $200 or
lost $100 until you make your decision concerning a second, identical gamble:
50% chance to win $200 and 50% chance to lose $100. Would you?:
(a) accept the second gamble.
(b) reject the second gamble.

36%
64%

The data show that a majority of subjects accepted the second gamble after
having won the first gamble and a majority also accepted the second gamble after
having lost the first gamble. However, the majority of subjects rejected the
second gamble when the outcome of the first was not known. An examination of
individual choices reveals that approximately 40% of the subjects accepted the
second gamble both after a gain in the first and after a loss. Among these,
however, 65% rejected the second gamble in the disjunctive condition, when the
outcome of the first gamble was not known. Indeed, this response pattern
(accepting in both conditions but rejecting in the disjunction) was the single most
frequent pattern, exhibited by 27% of all subjects. This pattern, which violates
Savage's (1954) sure-thing principle, cannot be attributed to unreliability
(Tversky & Shafir, 1992a).
The students above were offered a gamble with a positive expected value, and
an even chance of a non-trivial loss. Different reasons were likely to arise for
accepting the second gamble depending on the outcome of the first. In the win
condition, the decision maker is already up $200, so even a loss on the second
gamble leaves him or her ahead overall, which makes this option quite attractive.
In the lose condition, on the other hand, the decision maker is down $100.
Playing the second gamble offers a chance to "get out of the red", which for many
is more attractive than accepting a sure $100 loss. In the disjunctive condition,
however, the decision maker does not know whether she is up $200 or down $100;

30

Shafir et al. I Cognition 49 (1993) 11-36

she does not know, in other words, whether her reason for playing the second
gamble is that it is a no-loss proposition or, instead, that it provides a chance to
escape a sure loss. In the absence of a definite reason, fewer subjects accept the
second gamble.
This interpretation is further supported by the following modification of the
above problem, in which both outcomes of the first gamble were increased by
$400 so that the decision maker could not lose in either case.
Imagine that you have just played a game of chance that gave you a 50%
chance to win $600 and a 50% chance to win $300. Imagine that the coin has
already been tossed, but that you will not know whether you have won $600 or
$300 until you make your decision concerning a second gamble: 50% chance to
win $200 and 50% chance to lose $100.
A total of 171 subjects were presented with this problem, equally divided into
three groups. One group was told that they had won $300 on the first gamble, a
second group was told that they had won $600 on the first gamble, and the third
group was told that the outcome of the first gamble - $300 or $600 - was not
known (the disjunctive version). In all cases, subjects had to decide whether to
accept or to reject the second gamble which, as in the previous problem, consisted
of an even chance to win $200 or lose $100. The percentage of subjects who
accepted the second gamble in the $300, $600, and disjunctive versions, were
69%, 75%, and 73%, respectively. (Recall that the corresponding figures for the
original problem were 59%, 69%, and 36%; essentially identical figures were
obtained in a between-subjects replication of that problem.) In contrast to the
original problem, the second gamble in this modified problem was equally popular
in the disjunctive as in the non-disjunctive versions. Whereas in the original
scenario the second gamble amounted to either a no-loss proposition or a chance
to avoid a sure loss, in the modified scenario the second gamble amounts to a
no-loss proposition regardless of the outcome of the first gamble. The increased
popularity of the second gamble in the modified problem shows that it is not the
disjunctive situation itself that discourages people from playing. Rather, it is the
lack of a specific reason that seems to drive the effect: when the same reason
applies regardless of outcome, the disjunction no longer reduces the tendency to
accept the gamble.
As illustrated above, changes in the context of decision are likely to alter the
reasons that subjects bring to mind and, consequently, their choices. Elsewhere
(Shafir & Tversky, 1992) we describe a disjunction effect in the context of a
one-shot prisoner's dilemma game, played on a computer for real payoffs.
Subjects (n = 80) played a series of prisoner's dilemma games, without feedback,
each against a different unknown player. In this setup, the rate of cooperation
was 3% when subjects knew that the other player had defected, and 16% when
they knew that the other'had cooperated. However, when subjects did not know

Skafir et at. I Cognition 49 (1993) 11-36

31

whether the other player had cooperated or defected (the standard version of the
prisoner's dilemma game) the rate of cooperation rose to 37%. Thus, many
subjects defected when they knew the other's choice - be it cooperation or
defection - but cooperated when the other player's choice was not known. Shark
and Tversky (1992) attribute this pattern to the different perspectives that
underlie subjects' behavior under uncertainty as opposed to when the uncertainty
is resolved. In particular, we suggest that the reasons for competing are more
compelling when the other player's decision is known and the payoff depends on
the subject alone, than when the other's chosen strategy is uncertain, and the
outcome of the game depends on the choices of both players.
The above "disjunctive" manipulation - which has no direct bearing from the
point of view of value maximization - appears to influence the reasons for
decision that people bring to mind. Another kind of manipulation that seems to
alter people's reasons without bearing directly on options' values is described in
what follows.

6. Non-valued features

Reasons for choice or rejection often refer to specific features of the options
under consideration. The positive features of an option typically provide reasons
for choosing that option and its negative features typically provide reasons for
rejection. What happens when we add features that are neither attractive nor
aversive? Can choice be influenced by features that have little or no value?
Simonson and his colleagues have conducted a number of studies on the effects
of non-valued features, and tested the hypothesis that people are reluctant to
choose alternatives that are supported by reasons that they do not find appealing.
In one study, for example, Simonson, Nowlis, and Simonson (in press) predicted
that people would be less likely to choose an alternative that was chosen by
another person for a reason that does not apply to them. UC Berkeley business
students (n = 113) were told that, because of budget cuts and in order to save
paper and duplicating costs, a questionnaire that they will receive was designed
for use by two respondents. Thus, when subjects had to enter a choice, they could
see the choice made by the previous "respondent" and the reason given for it.
The choices and reasons of the previous respondents were systematically manipulated. One problem, for example, offered a choice between attending the MBA
programs at Northwestern and UCLA. In one version of the questionnaire, the
previous respondent had selected Northwestern, and provided the (handwritten)
reason, "I have many relatives in the Chicago area." Because this reason does not
apply to most subjects, it was expected to reduce their likelihood of choosing
Northwestern. In a second version, no reason was given for the choice of
Northwestern. As expected, those exposed to an irrelevant reason were less likely

32

Shafir et αϊ. I Cognition 49 (1993) 11-36

to choose Northwestern than subjects who saw the other respondent's choice but
not his or her reason (23% vs. 43%, p<.05). It should be noted that both
Northwestern and UCLA are well known to most subjects (Northwestern
currently has the highest ranked MBA program; the UCLA program is ranked
high and belongs to the same UC system as Berkeley). Thus, it is unlikely that
subjects made inferences about the quality of Northwestern based on the fact that
another respondent chose it because he or she had relatives in Chicago.
In a related study, Simonson, Carmon, and O'Curry (in press) showed that
endowing an option with a feature that was intended to be positive but, in fact,
has no value for the decision maker can reduce the tendency to choose that
option, even when subjects realize that they are not paying for the added feature.
For example, an offer to purchase a collector's plate - that most did not want - if
one buys a particular brand of cake mix was shown to lower the tendency to buy
that particular brand relative to a second, comparable cake mix brand (from 31%
to 14%, ρ < .05). Choosing brands that offer worthless bonuses was judged (in a
related study) as more difficult to justify and as more susceptible to criticism. An
analysis of verbal protocols showed that a majority of those who failed to select
the endowed option explicitly mentioned not needing the added feature. It should
be noted that sale promotions, such as the one involving the collector's plate offer
above, are currently employed by a wide range of companies and there is no
evidence that they lead to any inferences about the quality of the promoted
product (e.g., Blattberg & Neslin, 1990).
The above manipulations all added "positive", albeit weak or irrelevant,
features, which should not diminish an option's value; yet, they apparently
provide a reason against choosing the option, especially when other options are
otherwise equally attractive. Evidently, the addition of a potentially attractive
feature that proves useless can provide a reason to reject the option in favor of a
competing alternative that has no "wasted" features.

7. Concluding remarks
People's choices may occasionally stem from affective judgments that preclude
a thorough evaluation of the options (cf. Zajonc, 1980). In such cases, an analysis
of the reasons for choice may prove unwarranted and, when attempted by the
decision maker, may actually result in a different, and possibly inferior, decision
(Wilson & Schooler, 1991). Other choices, furthermore, may follow standard
operating procedures that involve minimal reflective effort. Many decisions,
nonetheless, result from a careful evaluation of options, in which people attempt
to arrive at what they believe is the best choice. Having discarded the less
attractive options and faced with a choice that is hard to resolve, people often
search for a compelling rationale for choosing one alternative over another. In

Shafir et al. I Cognition 49 (1993) 11-36

33

this paper, we presented an analysis of the role of reasons in decision making, and
considered ways in which an analysis based on reasons may contribute to the
standard quantitative approach based on the maximization of value. A number of
hypotheses that derive from this perspective were investigated in experimental
settings.
The reasons that enter into the making of decisions are likely to be intricate
and diverse. In the preceding sections we have attempted to identify a few general
principles that govern the role of reasons in decision making, and thus some of
the fundamental ways in which thinking about reasons is likely to contribute to
our understanding of the making of decisions. A reliance on the more important
dimensions - those likely to provide more compelling reasons for choice - was
shown in section 1 to predict preferences between previously equated options.
The notions of compatibility and salience were summoned in section 2 to account
for the differential weighting of reasons in a choice versus rejection task.
Reasons, it appears, lend themselves to certain framing manipulations that are
harder to explain from the perspective of value maximization. In section 3,
manipulating the precise relationships between competing alternatives was shown
to enhance or reduce conflict, yielding decisions that were easier or more difficult
to rationalize and justify. Providing a context that presents compelling reasons for
choosing an option apparently increases people's tendency to opt for that option,
whereas comparing alternatives that render the aforementioned reasons less
compelling tends to increase people's tendency to maintain the status quo or
search for other alternatives. The ability of the context of decision to generate
reasons that affect choice was further discussed in section 4, where the addition
and removal of competing alternatives was interpreted as generating arguments
for choice based on comparative considerations of relative advantages and
compromise. The relative weakness of disjunctive reasons was discussed in section
5. There, a number of studies contrasted people's willingness to reach a decision
based on a definite reason for choice, with their reluctance to arrive at a decision
in the presence of uncertainty about which reason is actually relevant to the case
at hand. Section 6 briefly reviewed choice situations in which the addition of
purported reasons for choosing an option, which subjects did not find compelling,
was seen to diminish their tendency to opt for that option, even though its value
had not diminished.
The nature of the reasons that guide decision, and the ways in which they
interact, await further investigation. There is evidence to suggest that a wide
variety of arguments play a role in decision making. We often search for a
convincing rationale for the decisions that we make, whether for inter-personal
purposes, so that we can explain to others the reasons for our decision, or for
intra-personal motives, so that we may feel confident of having made the "right"
choice. Attitudes toward risk and loss can sometimes be rationalized on the basis
of common myths or cliches, and choices are sometimes made on the basis of

34

Shafir et al. I Cognition 49 (1993) 11-36

moral or prudential principles that are used to override specific cost-benefit
calculations (cf. Prelec & Herrnstein, 1991). Formal decision rules, moreover,
may sometimes act as arguments in people's deliberations. Thus, when choosing
between options χ and z, we may realize that, sometime earlier, we had preferred
χ over у and у over z and that, therefore, by transitivity, we should now choose χ
over z. Montgomery (1983) has argued that people look for dominance structures
in decision problems because they provide a compelling reason for choice._
Similarly, Tversky and Shafir (1992a) have shown that detecting the applicability
of the sure-thing principle to a decision situation leads people to act in accord
with this principle's compelling rationale. Indeed, it has been repeatedly observed
that the axioms of rational choice which are often violated in non-transparent
situations are generally satisfied when their application is transparent (e.g.,
Tversky & Kahneman, 1986). These results suggest that the axioms of rational
choice act as compelling arguments, or reasons, for making a particular decision
when their applicability has been detected, not as universal laws that constrain
people's choices.
In contrast to the classical theory that assumes stable values and preferences, it
appears that people often do not have well-established values, and that preferences are actually constructed - not merely revealed - during their elicitation
(cf. Payne, Bettman, & Johnson, 1992). A reason-based approach lends itself well
to such a constructive interpretation. Decisions, according to this analysis, are
often reached by focusing on reasons that justify the selection of one option over
another. Different frames, contexts, and elicitation procedures highlight different
aspects of the options and bring forth different reasons and considerations that
influence decision.
The reliance on reasons to explain experimental findings has been the hallmark
of social psychological analyses. Accounts of dissonance (Wicklund & Brehm,
1976) and self-perception (Bern, 1972), for example, focus on the reasons that
people muster in an attempt to explain their counter-attitudinal behaviors.
Similarly, attribution theory (Heider, 1980) centers around the reasons that
people attribute to others' behavior. These studies, however, have primarily
focused on postdecisional rationalization rather than predecisional conflict.
Although the two processes are closely related, there are nevertheless some
important differences. Much of the work in social psychology has investigated
how people's decisions affect the way they think. The present paper, in contrast,
has considered how the reasons that enter into people's thinking about a problem
influence their decision. A number of researchers have recently begun to explore
related issues. Billig (1987), for example, has adopted a rhetorical approach to
understanding social psychological issues, according to which "our inner deliberations are silent arguments conducted within a single self (p. 5). Related
"explanation-based" models of decision making have been applied by Pennington
and Hastie (1988, 1992) to account for judicial decisions, and the importance of

Shafir et al. I Cognition 49 (1993) 11-36

35

social accountability in choice has been addressed by Tetlock (1992). From a
philosophical perspective, a recent essay by Schick (1991) analyzes various
decisions from the point of view of practical reason. An influential earlier work is
Toulmin's (1950) study of the role of arguments in ethical reasoning.
In this article, we have attempted to explore some of the ways in which reasons
and arguments enter into people's decisions. A reason-based analysis may come
closer to capturing part of the psychology that underlies decision and thus may
help shed light on a number of phenomena that remain counterintuitive from the
perspective of the classical theory. It is instructive to note that many of the
experimental studies described in this paper were motivated by intuitions
stemming from a qualitative analysis based on reasons, not from a value-based
perspective, even if they can later be interpreted in that fashion. We do not
propose that accounts based on reasons replace value-based models of choice.
Rather, we suggest that an analysis of reasons may illuminate some aspects of
reflective choice, and generate new hypotheses for further study.

References
Allison, G.T. (1971). Essence of decision: explaining the Cuban missile crisis. Boston: Little Brown.
Bern, D.J. (1972). Self-perception theory. In L. Berkowitz (Ed.). Advances in experimental social
psychology (Vol. 6). New York: Academic Press.
Berman, L. (1982). Planning a tragedy. New York: Norton.
Belts, R., & Gelb, L. (1979). The irony of Vietnam: the system worked. Washington, DC: Brookings
Institution.
Bigelow, J. (Ed.) (1887). The complete works of Benjamin Franklin (Vol. 4). New York: Putnam.
Billig, M. (1987). Arguing and thinking: a rhetorical approach to social psychology. New York:
Cambridge University Press.
Blattberg, R.C., & Neslin, S.A. (1990). Sales promotion: concepts, methods, and strategies.
Englewood Cliffs, NJ: Prentice-Hall.
Heider, F. (1980). The psychology of interpersonal relations. New York: Wiley.
Huber, J., Payne, J.W.. and Puto, C. (1982). Adding asymmetrically dominated alternatives:
violations of regularity and the similarity hypothesis. Journal of Consumer Research, 9, 90-98.
James, W. (1981). The principles of psychology (Vol. 2). Cambridge, MA: Harvard University Press.
Kahneman, D., & Tversky, A. (1979). Prospect theory: an analysis of decision under risk.
Econometrica, 47, 263-291.
Montgomery, H. (1983). Decision rules and the search for a dominance structure: towards a process
model of decision making. In P. Humphreys, O. Svenson, & A. Vari (Eds.), Analyzing and
aiding decision processes. Amsterdam: North-Holland.
Nisbett, R.E., & Wilson, T.D. (1977). Telling more than we can know: verbal reports on mental
processes. Psychological Review, 84, 231-259.
Payne, J.W., Bettman, J.R., & Johnson, E.J. (1992). Behavioral decision research: a constructive
process perspective. Annual Review of Psychology, 43, 87-131.
Pennington, N.. & Hastie, R. (1988). Explanation-based decision making: effects of memory structure
on judgment. Journal of Experimental Psychology: Learning, Memory, and Cognition, 14,
521-533.
Pennington,"N., & Hastie, R. (1992). Explaining the evidence: tests of the story model for juror
decision making. Journal of Personality and Social Psychology, 62, 189-206.

36

Shafir et al. I Cognition 49 (1993) 11-36

- Prelec, D., & Herrnstein, R.J. (1991). Preferences or principles: alternative guidelines for choice. In
R.J. Zeckhauser (Ed.), Strategy and choice. Cambridge, MA: MIT Press.
Redelmeier, D., & Shafir, E. (1993). Medical decisions over multiple alternatives. Working paper,
University of Toronto.
Savage, L.J. (1954). The foundations of statistics. New York: Wiley.
Schick, F. (1991). Understanding action: an essay on reasons. New York: Cambridge University Press.
Shafer, G. (1986). Savage revisited. Statistical Science, 1, 463-485.
» Shafir, E. (1993). Choosing versus rejecting: why some options are both better and worse than others.
Memory & Cognition, 21, 546-556.
- Shafir, E., & Tversky, A. (1992). Thinking through uncertainty: nonconsequentia! reasoning and
choice. Cognitive Psychology, 24, 449-474.
Simonson, I. (1989). Choice based on reasons: the case of attraction and compromise effects. Journal
of Consumer Research, 16, 158-174.
Simonson, I., Carmon, Z., & O'Curry, S. (in press). Experimental evidence on the negative effect of
unique product features and sales promotions on brand choice. Marketing Science.
Simonson, I., Nowlis, S., & Simonson. Y. (in press). The effect of irrelevant preference arguments on
consumer choice. Journal of Consumer Psychology.
. Simonson, I., & Tversky, A. (1992). Choice in context: tradeoff contrast and extremeness aversion.
Journal of Marketing Research, 29, 281-295.
Slovic, P. (1975). Choice between equally valued alternatives. Journal of Experimental Psychology:
Human Perception and Performance, I, 280-287.
- Slovic, P. (1990). Choice. In D. Osherson & E. Smith (Eds.), An invitation to cognitive science (Vol.
3). Cambridge, MA: MIT Press.
Telhami. S. (1990). Power and leadership in international bargaining: the path to the Camp David
accords. New York: Columbia University Press.
Tetlock, P.E. (1992). The impact of accountability on judgment and choice: toward a social
contingency model. In M.P. Zanna (Ed.), Advances in experimental social psychology (Vol.
25). New York: Academic Press.
Toulmin, S. (1950). The place of reason in ethics. New York: Cambridge University Press.
Tversky, A., & Kahneman, D. (1986). Rational choice and the framing of decisions. Journal of
Business, 59, 251-278.
Tversky, A., Sattath, S., & Slovic, P. (1988). Contingent weighting in judgment and choice.
Psychological Review, 95, 371-384.
Tversky, A., & Shafir, E. (1992a). The disjunction effect in choice under uncertainty. Psychological
Science, 3, 305-309.
Tversky, A., & Shafir, E. (1992b). Choice under conflict: the dynamics of deferred decision.
"" Psychological Science, 3, 358-361.
,
Tversky, A., & Simonson, I. (in press),, Context-dependent preferences. Management Science. ·"'' ·von Neumann, J., & Morgenstern, O. (1947). Theory of games and economic behavior. Princeton, NJ:
Princeton University Press.
, Wedell, D.H. (1991). Distinguishing among models of contextually induced preference reversals.
Journal of Experimental Psychology: Learning, Memory, and Cognition, 17, 767-778.
Wicklund, R.A., & Brehm, J.W. (1976). Perspectives on cognitive dissonance. Hillsdale, NJ: Erlbaum.
Wilson, T.D., & Schooler, J.W. (1991). Thinking too much: introspection can reduce the quality of
preferences and decisions. Journal of Personality and Social Psychology, 60, 181-192.
Zajonc, B. (1980). Preferences without inferences. American Psychologist, 35, 151-175.