Explanations Versus Applications: The Explanatory Power of Valuable Beliefs

Abstract

People hold beliefs that vary not only in their perceived truth, but also in their value to the believer--their meaning, relevance, and importance. We argue that a belief's value is determined, at least in part, by its explanatory power. Highly valuable beliefs are those that can uniquely explain and organize a diverse set of observations. Less valuable beliefs, in contrast, are those that can be explained by other observations, or that explain and organize few observations. The results of three experiments are consistent with these hypotheses. These experiments demonstrate that applying either scientific or religious beliefs to explain other observations increases the perceived value of those beliefs, whereas generating explanations for the existence of beliefs decreases their perceived value. Discussion focuses on the implications of these findings for people's resistance to explaining their own beliefs, for the perceived value of science and religion, and for culture wars between people holding opposing beliefs.

Full text

Research Article
Explanations Versus
Applications
The Explanatory Power of Valuable Beliefs
Jesse Preston
1
and Nicholas Epley
2
1
Harvard University and
2
University of Chicago
ABSTRACT—People hold beliefs that vary not only in their
perceived truth, but also in their value to the believer—
their meaning, relevance, and importance. We argue that
a belief’s value is determined, at least in part, by its ex-
planatory power. Highly valuable beliefs are those that
can uniquely explain and organize a diverse set of observa-
tions. Less valuable beliefs, in contrast, are those that can
be explained by other observations, or that explain and
organize few observations. The results of three experi-
ments are consistent with these hypotheses. These experi-
ments demonstrate that applying either scientific or
religious beliefs to explain other observations increases the
perceived value of those beliefs, whereas generating ex-
planations for the existence of beliefs decreases their per-
ceived value. Discussion focuses on the implications of
these findings for people’s resistance to explaining their
own beliefs, for the perceived value of science and religion,
and for culture wars between people holding opposing
beliefs.
Beliefs are propositions held to be true, and the average person
holds more beliefs than anyone would care to count. But not all
of these beliefs are equally valuable. Some—such as belief in
God—are vigorously defended when called into question,
whereas others—such as the belief that it will rain tomorrow—
are not. And some—such as those of Democrats versus Re-
publicans—create intense cultural conflicts between believers,
whereas others—such as those of dog lovers versus cat lovers—
do not. Valuable beliefs are those that are personally meaning-
ful, relevant, and important to people in their daily lives, and the
research we report here investigates one important mechanism
by which beliefs become valuable.
To be sure, beliefs are valued for a variety of reasons: for emo-
tional comfort (Lerner, 1980), self-expression (Prentice, 1987), ego
defense (Katz, 1960), and behavior regulation (Greenwald, 1989),
among others (Eagly & Chaiken, 1998). Perhaps a belief’s most
basic instrumental function, however, is to serve as an explanation
for one’s observations. Belief in free will, for instance, explains
one’s own and other people’s actions (Wegner, 2002). Belief in
right-wing conspiracies explains presidential impeachments. And
religious beliefs explain the origin of the universe and life after
death. Many of the beliefs people possess are in some sense causal
explanations that organize their observations and reduce com-
plexity (e.g., life exists because of God), thereby providing ex-
pectations for the future (Berlyne, 1960; Gilbert, 1991; Heider,
1958) and reducing the anxiety associated with uncertainty (Or-
tony, Clore, & Collins, 1988). To the extent that beliefs serve as
explanations for one’s observations, their value should be a func-
tion of their explanatory power.
The idea that beliefs serve as explanations is certainly not new
(e.g., Allport, 1935; Frazer, 1890/1923; Thagard, 1989). How-
ever, unlike previous functional accounts of belief, our account
suggests that it is not simply the perceived truth of a belief that is
influenced by its explanatory power, but its perceived value—its
meaning, importance, and personal relevance—as well. As
people apply a belief to explain more observations, the value of
that belief should increase. Applying a belief to other observa-
tions positions it as a first cause in a sequence of events, and
unites different effects together through a mutual cause. The
belief in love as a critical ingredient in romantic relationships,
for example, can explain a spouse’s steadfast monogamy, lifelong
devotion, and tender laughter at one’s bad jokes. With each new
application, belief in the importance, meaningfulness, and
personal relevance of love should increase. We therefore predict
that applying a belief to explain one’s observations should
increase its perceived value.
Address correspondence to Nicholas Epley, University of Chicago,
5807 S. Woodlawn Ave., Chicago, IL 60637, e-mail:epley@chicagogsb.
edu, or to Jesse Preston, e-mail: jpreston@post.harvard.edu.
PSYCH O L O G I C A L S C I E N C E
826 Volume 16—Number 10Copyright r2005 American Psychological Society

Perhaps more important, the flip side of this argument is that
explanatory power diminishes if the belief is itself explained by
other beliefs. Love for one’s spouse, for instance, can be ex-
plained by physical attraction, perceived similarity, and the
formal commitment that comes with marriage. Now all of the
beliefs that could previously be explained by love can be ex-
plained by other, more basic, beliefs. Thus, explaining a belief
also positions it in a causal sequence, but relegates the focal
belief to a secondary or mediating status that may make the focal
belief seem less important, meaningful, and relevant (Penne-
baker, 1990, 1997; Wilson, Gilbert, & Centerbar, 2003; Wort-
man, Silver, & Kessler, 1993). We therefore predict that
explaining a belief will cause it to lose some of its unique ex-
planatory power, and therefore lose some of its meaning, im-
portance, and relevance as well.
Notice that this example about love refers to the perceived
importance of love in a relationship, not to whether or not one is
actually loved in one’s own relationship, thus demonstrating that
the perceived value of a belief can be quite independent of its
perceived truth. A person may hold two beliefs to be equally
true, but can still value one belief over the other. In addition,
both applications of a belief and explanations for a belief can
provide evidence consistent with the validity of a belief. Indeed,
previous research has found that both explanations and appli-
cations of a belief can increase the extent to which that belief is
perceived to be a ‘‘good’’ explanation (Read & Marcus-Newhall,
1993). Our predictions about the perceived value of a belief,
then, are not dependent on altering the perceived truth or va-
lidity of a belief.
We tested our hypotheses in three experiments. In each,
participants considered either a novel or an existing belief and
were asked to focus on either applications of that belief (i.e.,
observations that the belief could explain) or explanations for
that belief (i.e., observations or underlying causes that could
explain the existence of the belief). We predicted that partici-
pants asked to apply beliefs would find them to be more valu-
able—that is, more meaningful, important, and personally
relevant—than participants asked to explain beliefs.
STUDY 1: CREATING BELIEFS
Participants in Study 1 were presented with one of two novel
scientific beliefs that are familiar to most psychologists but would
be considerably less so to the participants: (a) that people prefer
similarity in relationship partners or (b) that people with high self-
esteem are more likely to be aggressive than people with low self-
esteem. Participants in the applications condition were then asked
to apply their provided belief to other observations (i.e., to think of
observations that their belief could explain). Participants in the
explanations condition, in contrast, were asked to think of obser-
vations that could explain the belief (i.e., why people prefer sim-
ilarity, or why self-esteem might be linked to aggression).
Participants in a control condition neither applied nor explained
the provided belief. We expected that participants in the appli-
cations condition, compared with those in the control condition,
would rate their assigned belief as more valuable—more impor-
tant, meaningful, and relevant to society—whereas participants in
the explanations condition would rate their belief as less valuable
than would those in the control condition.
Study 1 also tested our secondary prediction that the per-
ceived value of a belief can vary somewhat independently of its
perceived truth. We tested this prediction by asking participants
to indicate the likelihood that the target belief was correct.
Method
Interested travelers in a Boston, MA, train station (N5171)
received a questionnaire describing one of two beliefs. One
group (n573) read about the widely documented relationship
between similarity and attraction (Berscheid & Reis, 1998):
Psychologists have argued that, whether choosing friends or fall-
ing in love, we are most attracted to people whose traits are similar
to our own. There seems to be wisdom in the old saying, ‘‘Birds of a
feather flock together.’’ (Myers, 1994, p. 18)
Another group (n598) read about the documented rela-
tionship between self-esteem and aggression (Baumeister,
Smart, & Boden, 1996):
Although intuition suggests that people who are depressed or low
in self-esteem are more likely to be violent or aggressive towards
others, some research demonstrates exactly the opposite. In fact,
people who are high in self-esteem are more likely to be aggressive
toward other people.
All participants were then randomly assigned to one of three
conditions. Those in the applications condition were asked to list
as many ‘‘implications or observations that this research finding
would explain.’’ Those in the explanations condition were asked to
list as many reasons ‘‘why this finding could come about.’’ Par-
ticipants in the control condition received no writing instructions.
All participants then reported the likelihood that the finding
was correct, using a scale ranging from 0% to 100%. Participants
then rated the perceived value of the finding. Specifically, they
indicated how important, meaningful, and personally relevant the
finding appeared to them, as well as how likely the finding was to
have an impact on society. All value ratings were made on 11-
point scales ranging from 0 (not at all) to 10 (a great deal). Finally,
participants rated the difficulty of the task on an 11-point scale
ranging from 0 (extremely easy) to 10 (extremely difficult).
Results and Discussion
The difficulty people experience when generating information is
often used as a cue for its validity (e.g., Schwarz, 1998), but
Volume 16—Number 10 827
Jesse Preston and Nicholas Epley

between-condition differences in difficulty cannot account for
our predicted effects as the difficulty ratings were inconsistent
with our predictions. A one-way analysis of variance (ANOVA)
on perceived difficulty yielded a marginally significant effect of
condition, F(2, 172) 52.82, p5.06, d50.27, with participants
in the control condition (who were asked to do the least) per-
ceiving the task as less difficult (M51.43) than those in the
applications (M52.54) or explanations (M52.12) condition.
These latter two conditions did not differ from each other, t<1.
To test our main hypothesis about explanatory power, we cre-
ated a single composite measure by averaging across the four
dependent measures (a5.85). As predicted, a 2 (scenario:
similarity vs. self-esteem) !3 (condition: explanations vs. control
vs. applications) ANOVA on this composite measure yielded a
main effect for condition, F(2, 169) 58.99, p<.001, Z
2
5.10.
There was no main effect of the particular scenario and no in-
teraction between scenario and condition. As can be seen in Table
1, participants in the applications condition found their assigned
belief to be the most valuable, whereas those in the explanations
condition found their assigned belief to be the least valuable,
linear contrast F(1, 174) 519.71, p<.001, d50.67.
Perceived correctness of the belief did not show the same
pattern, also as predicted. As can be seen in Table 1, there were
no differences in the perceived correctness across conditions (F
<1). A 2 (scenario) !3 (condition) !2 (measure: correctness
vs. value) ANOVA conducted on standardized values of cor-
rectness and value yielded only the predicted three-way inter-
action, F(2, 165) 59.01, p<.001, Z
2
5.10.
1
STUDY 2: OTHER PEOPLE’S BELIEFS
Understanding other people’s thoughts and beliefs is a central
feature of nearly all social interaction, and such understanding
influences people’s behavior and attitudes toward one another.
To the extent that the explanatory utility of a belief is used as a
guide for determining the value of one’s own beliefs, so too
should it be used as a guide when making inferences about other
people’s beliefs. A long line of research, however, suggests that
such social judgments tend to be egocentrically biased (Nick-
erson, 1999), and people are therefore likely to use their own
perceived value of a belief as an intuitive guide to others’ per-
ceived value. In order to highlight the role of explanatory power
in predicting other people’s beliefs without possible contami-
nation from egocentric biases, we asked participants in Study 2
to make inferences about a belief that none were likely to find
even remotely valuable. Specifically, participants in Study 2
were asked to consider the extent to which the ancient Greeks
valued their belief in the mythological god Poseidon, the god of
the sea. As in Study 1, we expected that participants led to focus
on applications would rate the belief as most valuable, whereas
those led to focus on explanations would rate it as least valuable.
Method
Interested travelers (N547) in a Boston, MA, train station
received a short paragraph about Poseidon:
In Greek Mythology, each god or goddess governed a specific part
of the world, or represented a specific part of life that the ancient
Greeks experienced. The sea was believed to be the realm of the
god Poseidon.
Participants randomly assigned to the applications condition
were then asked to list observations in the daily life of the Greeks
that Poseidon could explain, whereas those assigned to the ex-
planations condition were asked to list ‘‘observations that the
average Greek citizen believed could explain Poseidon’s be-
havior.’’ Participants in a control condition received no writing
instructions.
TABLE 1
Ratings of Value and Correctness in Study 1 and Study 2
Condition
Predicted linear
contrast
Belief and measure Explanations Control Applications t d
Similarity and attraction
Correctness 62.4 62.0 61.3 "0.21 .05
Total value 4.64 5.13 5.93 2.3
n
.55
Self-esteem and aggression
Correctness 53.3 49.4 52.2 "0.19 .04
Total value 3.80 4.81 5.81 3.86
nnn
.76
Greeks’ belief in Poseidon
Total value 5.51 6.48 7.36 2.97
nn
.90
Note. Correctness was rated on a 101-point scale ranging from 0 to 100. Totalvalue is the average across ratings of value, which were
made on 11-point scales ranging from 0 (not at all) to 11 (extremely).
n
p<.05.
nn
p<. 01.
nnn
p<.001.
1
Degrees of freedom differ for analyses of correctness and value because some
participants failed to respond to the correctness item.
828 Volume 16—Number 10
Explanatory Power

On the next page, participants were asked how important
Poseidon was to the ancient Greeks, how relevant Poseidon was
to the ancient Greeks, and how meaningful Poseidon was to the
average Greek citizen, using separate 11-point scales ranging
from 0 (not at all) to 10 (a great deal).
Results and Discussion
Our main prediction was that participants in the applications
condition would rate the belief in Poseidon as most valuable to
the ancient Greeks and that participants in the explanations
condition would list Poseidon as least valuable. A one-way
ANOVA on the overall composite measure of value (a5.79)
confirmed this prediction, F(2, 44) 54.45, p<.05, d50.64
(see Table 1). The perceived value of Poseidon increased as
people considered applications of the belief, and decreased as
people considered explanations for the belief.
Although the results of Study 1 and Study 2 are consistent
with our hypotheses, in neither study did we try to influence
the perceived value of an existing belief, nor did either study
investigate beliefs that are generally considered to be especial-
ly valuable. In Study 3, we sought to do just that.
STUDY 3: CHERISHED BELIEFS
Few beliefs are more valuable to people than their religious
beliefs. Wars are fought in defense of such beliefs, communities
and nations are organized around religious institutions, and
personal identities are often defined by the presence or absence
of religious affiliations. Study 3 investigated whether people’s
religious beliefs could be influenced by highlighting the ex-
planatory power of their God concepts. Much as in Study 2,
participants were asked either to apply their concept of God to
explain other observations (applications condition) or to con-
sider observations that could explain God’s behavior (explana-
tions condition).
In addition, Study 3 manipulated explanatory utility by al-
tering not only the observations participants were led to con-
sider, but also the number of observations they were asked to
generate. All else being equal, the more observations a belief
can explain, the more valuable it should appear to be. Con-
versely, the more a belief can be readily explained by other
observations, the less valuable it should appear to be. To in-
vestigate whether such effects would be observed with people’s
own religious beliefs, we asked participants in Study 3 to gen-
erate either 3 or 10 applications that God could explain, or 3 or
10 observations that could explain God’s behavior. We predicted
that increasing the explanatory power of God would increase the
perceived value of participants’ religious beliefs.
Because the vast majority of participants held Judeo-Chris-
tian beliefs, many were likely to see explaining God as ex-
ceedingly difficult (and perhaps somewhat inappropriate). In
this religious tradition, God is perceived to operate autono-
mously without being influenced by the natural world, and God’s
behavior is therefore uncaused in the traditional sense. Con-
sistent with this possibility, a pretest (n533) measuring the
perceived difficulty of the four conditions revealed a significant
main effect of the kind of observations listed (explanations vs.
applications), but no effect of the number of observations listed.
Listing observations that could explain God’s behavior was
perceived to be more difficult (M56.61) than applying God as
an explanation for other observations (M54.75), F(1, 29) 5
4.71, p5.04. To the extent that God cannot be readily ex-
plained, the belief in God may be relatively immune to a de-
crease in value. Nevertheless, we retained these conditions to
maintain symmetry with the previous experiments.
Method
Eighty interested Harvard undergraduates received a ques-
tionnaire informing them that this study was investigating peo-
ple’s religious beliefs. Those in the applications condition were
then asked to list either 3 or 10 observations that God can ex-
plain, whereas those in the explanations condition were asked to
list either 3 or 10 observations that can explain God’s behavior.
Participants who considered themselves atheists were asked to
list observations that a believer would likely make. When fin-
ished, participants answered four questions about their religious
beliefs: ‘‘What is the general importance of God in your life?’’
‘‘How important is God to you on a daily basis?’’ ‘‘How confident
are you that God exists?’’ and ‘‘To what extent do you feel you
have a personal relationship with God?’’ All responses were
made on 11-point scales ranging from 0 (not at all) to 10 (ex-
tremely). One final question asked participants to rate their faith
in God compared with the faith of the average Harvard student,
on a scale ranging from "5 (much weaker) to 5 (much stronger).
Results and Discussion
Twenty-three of the 80 participants considered themselves
atheists, which approximates the ratio in the Harvard popula-
tion. We excluded these participants, but including them does
not alter the significance levels of any of the following analyses.
To test our hypotheses, we first converted responses to the final
comparative faith question from a scale from "5 to 15 to a scale
from 0 to 10 by adding 5 to each response, thereby matching the
scales for the other items. We then created a composite measure
of belief in God by averaging together all five items (a5.98). A
2 (number of observations: 3 vs. 10) !2 (condition: explanations
vs. applications) between-participants ANOVA on this com-
posite score revealed a significant main effect of condition, F(1,
53) 58.06, p<.01, Z
2
5.13, indicating greater belief in the
applications condition than in the explanations condition.
Neither the main effect of number nor the two-way interaction
between condition and number was significant.
The pattern of means shown in Figure 1 suggests that the
number of items listed amplified the effects of the applications
Volume 16—Number 10 829
Jesse Preston and Nicholas Epley

condition, resulting in a substantial increase in reported belief
in God among participants asked to list 10 observations that God
could explain. Indeed, a follow-up contrast indicated that belief
was significantly higher in this condition than in the others, t(53)
52.67, p5.01, d50.77. There was not a reciprocal decrease
in belief as the number of explanations increased. This finding is
not especially surprising as most of our participants did not
believe that God’s actions could be explained by much at all. In
fact, none of the participants asked to list 10 explanations were
able to do so, and participants in this condition listed only an
average of 4.2 explanations—barely more than in the 3-expla-
nations condition. There was no difference in belief in God
between the 10-explanations and the 3-explanations conditions,
it appears, simply because participants were unable to generate
10 explanations for God’s behavior, and the experimental ma-
nipulation was therefore unsuccessful. In fact, the difficulty of
explaining God’s behavior may partially account for the ex-
tremely high value of belief in God. God is easy to apply but
difficult to explain.
GENERAL DISCUSSION
Beliefs held with confidence may vary considerably in their
perceived value to the believer, and the results of these three
experiments suggest that one important component of a belief ’s
value is its explanatory power. Those beliefs that can be broadly
applied to explain a variety of observations are considered to be
more meaningful, important, and personally relevant than those
that can be applied more narrowly. Whether considering novel
beliefs, other people’s beliefs, or their own cherished religious
beliefs, participants found beliefs to be more valuable when they
were led to apply their beliefs as explanations for other obser-
vations. In contrast, those beliefs that can readily be explained
by other observations seem to lose some of their explanatory
power, and therefore their perceived value. In both Studies 1 and
2, participants who considered observations that could explain a
belief found that belief to be less valuable than those who did not
consider such explanations.
To be sure, explanatory power is not the only mechanism
through which beliefs derive their value. For instance, the
perceived difficulty of generating applications or explanations
may well moderate the perceived importance of the belief. Al-
though we did not find that perceived difficulty could explain the
results of the present experiments, substantial evidence sug-
gests that a controlled manipulation of perceived difficulty
would moderate the value of a belief (Schwarz, 1998). Also,
variability in the importance of functional versus symbolic
sources of value may moderate the importance of explanatory
power. Functional sources of value—like explanatory power—
may be especially important when people are motivated to
possess accurate beliefs, whereas symbolic sources of value
such as self-expression may be relatively more important when
people are motivated to possess socially desirable beliefs. The
present research is therefore not intended to supplant existing
functional accounts of beliefs, but rather to expand on them by
demonstrating an additional, and we believe critically impor-
tant, source of a belief’s value.
We believe that, in addition to shedding new light on the
determinants of a belief’s value, these results have important
implications for persuasion. Like the marketer who points out
the unique functionality of his or her favored gadget that can
both slice and dice, so should influence peddlers advertise the
unique and wide variety of observations that their favored be-
liefs can uniquely explain. Attempts to explain those beliefs, of
course, should be left to their opponents.
Although we know of no research in the persuasion literature
that has tested the persuasive appeal of applications versus
explanations, we cannot help noticing that our predictions often
seem confirmed in scientific discourse. People attempting to
praise one’s research often do so by highlighting its wide array of
applications, whereas those attempting to belittle one’s research
do so by highlighting the host of existing mechanisms that could
explain one’s findings. What is more, higher levels of analysis
within scientific discourse often appear—rightly or wrongly—to
lose some of their appeal with the arrival of more basic levels of
analyses. Broadly speaking, for example, sociology can be ex-
plained by the mechanisms of social psychology, social psy-
chology can be explained by the mechanisms of cognitive
psychology, and all perhaps eventually will be explained by
Fig. 1. Mean perceived value of belief in God as a function of experi-
mental condition and number of observations in Study 3.
830 Volume 16—Number 10
Explanatory Power

neuroscience. Part of the appeal, then, of these lower levels of
analysis may be due to their apparent ability to explain higher-
level phenomena. Of course, whether lower levels of analyses
really ought to devalue higher levels of analyses is a functional
issue that depends on what, exactly, one is trying to predict or
explain.
We also believe these experiments can help account for
people’s resistance to explanations for their cherished beliefs.
Those of religious faith, for example, seem threatened when
scientific explanations—such as evolution—are offered for
observations otherwise explained by religious concepts or when
psychological concepts are used to explain religious belief it-
self. Even if these explanations do not impinge on the core tenets
of a religious ideology, they may nevertheless seem to devalue
religious beliefs, and lead to an intense resistance to such ex-
planations. Indeed, the history of science and religion is replete
with examples of such resistance. In some cases, it may be so
intense that believers wish to avoid the search for underlying
explanations altogether. Senator William Proxmire, for example,
justified giving one of his ‘‘Golden Fleece Awards’’ to Ellen
Berscheid and Elaine Hatfield for their research on love by
stating, ‘‘I don’t want the answer. I believe that 200 million other
Americans want to leave some things in life a mystery’’ (cited in
Hatfield & Walster, 1978, p. viii). Explanations for cherished
beliefs can devalue those beliefs to such an extent that people
may prefer to stop further understanding altogether. The results
reported here may therefore shed explanatory light on culture
wars that are likely to develop between groups who hold op-
posing beliefs, and thereby join the growing body of research
investigating how cultural beliefs and practices arise from basic
psychological processes (Gilbert, Brown, Pinel, & Wilson, 2000;
Heath, Bell, & Sternberg, 2001; Lyons & Kashima, 2001;
Schaller & Crandall, 2004).
Finally, this research suggests that the ultimately valuable
belief (a) explains everything and (b) is explained by nothing.
Few beliefs can manage this feat, but those associated with
science and religion are the most common contenders. We think
it is no accident that Western theology has historically depicted
God as the ‘‘unmoved First Mover.’’ Both science and religion
seek primary causes that can explain higher-level observations,
albeit through different methods. It is of little surprise, given our
findings, that believers in science and believers in religion so
often come into direct conflict. What these experiments suggest
is that at least some of this conflict can be attributed to the
psychological mechanisms that create valuable beliefs. What
these valuable beliefs share, our research suggests, is not simply
their perceived truth, but their power as explanations.
Acknowledgments—This research was supported by grants
from the National Science Foundation (SES0241544) and the
James S. Kemper Foundation Faculty Research Fund awarded
to Nicholas Epley. We thank Andrew Fichte, Samantha Frank-
lin, Nicholas Josefowitz, Richard Lonsdorf, Erin Rapien, Aram
Seo, and Audrey Tse for assistance conducting these experi-
ments; Daniel Wegner for advice during the course of this re-
search; and Richard Eibach, Chip Heath, and Zakary Tormala
for helpful comments on a previous draft.
REFERE NC ES
Allport, G.W. (1935). Attitudes. In C.A. Murchinson (Ed.), A handbook
of social psychology (Vol. 2, pp. 798–844). Worcester, MA: Clark
University Press.
Baumeister, R.F., Smart, L., & Boden, J.M. (1996). Relation of threat-
ened egotism to violence and aggression: The dark side of high
self-esteem. Psychological Review,103, 5–33.
Berlyne, D.E. (1960). Conflict, arousal, and curiosity. New York:
McGraw-Hill.
Berscheid, E., & Reis, H.T. (1998). Attraction and close relationships.
In D.T. Gilbert, S.T. Fiske, & G. Lindzey (Eds.), The handbook of
social psychology (4th ed., Vol. 2, pp. 193–281). New York:
McGraw-Hill.
Eagly, A.H., & Chaiken, S. (1998). Attitude structure and function. In
D.T. Gilbert, S.T. Fiske, & G. Lindzey (Eds.), The handbook of
social psychology (4th ed., Vol. 1, pp. 269–322). New York:
McGraw-Hill.
Frazer, J. (1923). The golden bough: A study in magic and religion
(abridged ed.). Oxford, England: Macmillan. (Original work pub-
lished 1890)
Gilbert, D.T. (1991). How mental systems believe. American Psychol-
ogist,46, 107–119.
Gilbert, D.T., Brown, R.P., Pinel, E.C., & Wilson, T.D. (2000). The il-
lusion of external agency. Journal of Personality and Social Psy-
chology,79, 690–700.
Greenwald, A.G. (1989). Why attitudes are important: Defining attitude
and attitude theory 20 years later. In A. Pratkanis, S. Breckler, &
A. Greenwald (Eds.), Attitude structure and function (pp. 429–
440). Hillsdale, NJ: Erlbaum.
Hatfield, E., & Walster, G.W. (1978). A new look at love. Lanham, MD:
University Press of America.
Heath, C., Bell, C., & Sternberg, E. (2001). Emotional selection in
memes: The case of urban legends. Journal of Personality and
Social Psychology,81, 1028–1041.
Heider, F. (1958). The psychology of interpersonal relations. New York:
Wiley.
Katz, D. (1960). The functional approach to the study of attitudes.
Public Opinion Quarterly,24, 163–204.
Lerner, M.J. (1980). The belief in a just world: A fundamental delusion.
New York: Plenum.
Lyons, A., & Kashima, Y. (2001). The reproduction of culture: Com-
munication processes tend to maintain cultural stereotypes. Social
Cognition,19, 372–394.
Myers, D.G. (1994). Exploring social psychology. New York: McGraw-Hill.
Nickerson, R.S. (1999). How we know—and sometimes misjudge—
what others know: Imputing one’s own knowledge to others. Psy-
chological Bulletin,125, 737–760.
Ortony, A., Clore, G.L., & Collins, A. (1988). The cognitive structure of
emotions. Cambridge, England: Cambridge University Press.
Pennebaker, J.W. (1990). Opening up: The healing power of expressing
emotions. New York: Guilford.
Pennebaker, J.W. (1997). Writing about emotional experiences as a
therapeutic process. Psychological Science,8, 162–166.
Volume 16—Number 10 831
Jesse Preston and Nicholas Epley

Prentice, D.A. (1987). Psychological correspondence of possessions,
attitudes, and values. Journal of Personality and Social Psychol-
ogy,53, 993–1003.
Read, S.J., & Marcus-Newhall, A. (1993). Explanatory coherence in
social explanations: A parallel distributed processing account.
Journal of Personality and Social Psychology,65, 425–447.
Schaller, M., & Crandall, C.S. (2004). The psychological foundations of
culture. Mahwah, NJ: Erlbaum.
Schwarz, N. (1998). Accessible content and accessibility experiences:
The interplay of declarative and experiential information in
judgment. Personality and Social Psychology Review,2, 87–99.
Thagard, P. (1989). Explanatory coherence. Behavioral and Brain
Sciences,12, 435–467.
Wegner, D.W. (2002). The illusion of conscious will. Cambridge, MA:
MIT Press.
Wilson, T.D., Gilbert, D.T., & Centerbar, D.B. (2003). Making sense:
The causes of emotional evanescence. In I. Brocas & J. Carrillo
(Eds.), The psychology of economic decisions: Vol. 1. Rationality
and well being (pp. 209–233). New York: Oxford University Press.
Wortman, C.B., Silver, R.C., & Kessler, R.C. (1993). The meaning of
loss and adjustment to bereavement. In M.S. Stroebe, W. Stroebe,
& R.O. Hansson (Eds.), Handbook of bereavement: Theory, re-
search, and intervention (pp. 349–366). New York: Cambridge
University Press.
(RECEIVED 9/1/04; REVISION ACCEPTED 11/23/04;
FINAL MATERIALS RECEIVED 11/29/04)
832 Volume 16—Number 10
Explanatory Power