Exaggerated, Mispredicted, and Misplaced: When "It's the Thought That Counts" in Gift Exchanges

Abstract

Gift-giving involves both the objective value of a gift and the symbolic meaning of the exchange. The objective value is sometimes considered of secondary importance as when people claim, "It's the thought that counts." We evaluated when and how mental state inferences count in gift exchanges. Because considering another's thoughts requires motivation and deliberation, we predicted gift givers' thoughts would increase receivers' appreciation only when triggered to consider a giver's thoughts, such as when a friend gives a bad gift. Because gift givers do not experience this trigger, we expected they would mispredict when their thoughts count and when they do not. Three experiments support these predictions. A final experiment demonstrated that thoughts "count" for givers by increasing social connection to the receiver. These results suggest that mental state inferences are not automatic in social interactions and that inferences about how much thoughts count are systematically miscalibrated. (PsycINFO Database Record (c) 2012 APA, all rights reserved).

Full text

Exaggerated, Mispredicted, and Misplaced:
When “It’s the Thought That Counts” in Gift Exchanges
Yan Zhang
National University of Singapore Nicholas Epley
University of Chicago
Gift-giving involves both the objective value of a gift and the symbolic meaning of the exchange. The
objective value is sometimes considered of secondary importance as when people claim, “It’s the thought
that counts.” We evaluated when and how mental state inferences count in gift exchanges. Because
considering another’s thoughts requires motivation and deliberation, we predicted gift givers’ thoughts
would increase receivers’ appreciation only when triggered to consider a giver’s thoughts, such as when
a friend gives a bad gift. Because gift givers do not experience this trigger, we expected they would
mispredict when their thoughts count and when they do not. Three experiments support these predictions.
A final experiment demonstrated that thoughts “count” for givers by increasing social connection to the
receiver. These results suggest that mental state inferences are not automatic in social interactions and
that inferences about how much thoughts count are systematically miscalibrated.
Keywords: theory of mind, mind perception, social exchange, social cognition, social connection
Supplemental materials: http://dx.doi.org/10.1037/a0029223.supp
When people exchange gifts, it is often said, “It’s the thought
that counts.” This common wisdom suggests that the appreciation
and gratitude that comes from receiving a gift are based on more
than a gift’s objective quality. It also depends on how much
thought was put into the gift. Gifts chosen thoughtfully, with time
and effort taken to identify the perfect gift, are presumably appre-
ciated more than gifts chosen thoughtlessly, such as by choosing a
gift randomly or from a receiver’s wish list. Not only does a gift
receiver value the gift itself, but also feels appreciation and grat-
itude for the thought that went into the gift as an independent
source of value. This research examined when, how, and for whom
a gift giver’s thoughts actually count in gift exchanges.
At first glance, the common wisdom that “thoughts count”
appears not only intuitively credible but also empirically reason-
able. Indeed, psychological research demonstrates that people
care a great deal about others’ thoughts and intentions, using
inferences about others’ mental states as a fundamental guide to
social judgments and behavior (Epley & Waytz, 2010). For exam-
ple, people appear more responsible for actions they perform
intentionally than accidentally (Alicke, 2000), are liked more when
they help with prosocial intentions than with selfish intentions
(Ames, Flynn, & Weber, 2004), appear more prejudiced when they
are intentionally sexist than unintentionally sexist (Swim, Scott,
Sechrist, Campbell, & Stangor, 2003), and are punished more
severely for intentional harm than for accidental harm (Hogue &
Pebbles, 1997; Kleinke, Wallis, & Stadler, 1992). Even being
shocked with electricity hurts more when done intentionally than
when done accidentally (Gray & Wegner, 2008). Others’ thoughts,
in many domains of social life, do indeed seem to count.
We suggest, however, that the psychological processes that lead
people to consider others’ mental states calls the common wisdom
on the importance of thoughts in gift exchanges into question. In
particular, we suggest the presumed importance of a gift giver’s
thoughts is miscalibrated in three ways: (1) It is exaggerated for
receivers, (2) it is mispredicted by givers, and (3) it is misplaced by
focusing only on gift receivers when a gift giver’s thoughts actu-
ally count reliably for the givers themselves.
Exaggerated Impact: Thoughts Count for Receivers
When Mental State Reasoning Is Activated
First, we predicted that a gift giver’s thoughts—the time, effort,
and care expended to choose an ideal gift—would count for gift
receivers by increasing their gratitude and appreciation, but only
when receivers are triggered to think about a giver’s thoughts.
Inferences about others’ intentions, motives, goals, or thought
processes cannot influence a person’s subsequent judgment, eval-
uations, or choices if those preceding mental state inferences have
never been made. Mental states are, after all, inherently invisible,
making them relatively easy to overlook. Although most human
beings have the capacity to reason about others’ mental states, we
argue that this capacity is not used unless it is activated implicitly
by the social context itself (e.g., Young & Saxe, 2009) or explicitly
This article was published Online First July 9, 2012.
Yan Zhang, Business School, National University of Singapore, Singa-
pore; Nicholas Epley, Center for Decision Research, Booth School of
Business, University of Chicago.
We thank the National University of Singapore, the University of
Chicago, and the Templeton Foundation for research support. We thank
Leaf Van Boven for suggesting the hypothesis tested in Experiment 4, Oleg
Urminsky for statistical advice, and Ashley Angulo, Eric Guo, Mina Kang,
Jasmine Kwong, Rachel Meng, Alyssa Pappas, Maimouna Thioune, and
Evan Weingarten for assistance conducting the experiments.
Correspondence concerning this article should be addressed to Nicholas
Epley, Booth School of Business, University of Chicago, 5807 South
Woodlawn Avenue, Chicago, IL 60637. E-mail: yan.zhang@nus.edu.sg or
epley@chicagobooth.edu
Journal of Experimental Psychology: General © 2012 American Psychological Association
2012, Vol. 141, No. 4, 667–681 0096-3445/12/$12.00 DOI: 10.1037/a0029223
667
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

by a direct request to consider another’s thoughts or perspective
(e.g., Back & Apperly, 2010; Epley, Caruso, & Bazerman, 2006;
Galinsky, Wang, & Ku, 2008). In contrast, the objective quality of
a gift is directly visible and likely to capture a gift receiver’s
attention immediately. Once capturing attention, the object’s over-
all quality—the extent to which it is positive or negative, liked or
disliked—tends to be evaluated automatically (Bargh, Chaiken,
Govender, & Pratto, 1992; Duckworth, Bargh, Garcia, & Chaiken,
2002; Fazio, 2001; Greenwald, Klinger, & Liu, 1989). We there-
fore predicted that the objective quality of the gift would be
primary in the receiver’s evaluations and that the thought a gift
giver expended would be of secondary importance. Gift receivers
would consider a gift giver’s thoughts only when triggered to go
beyond the automatic evaluation of the gift itself.
Existing research suggests that the process of thinking about the
minds of others is activated when these agents behave negatively
(Morewedge, 2009) or unexpectedly (Waytz et al., 2010), as part
of a more general attempt to explain the causes of an agent’s
behavior (Pittman & Pittman, 1980; Wong & Weiner, 1981). In
gift exchanges, the clearest unexpected negative outcome occurs
when a close other gives a bad gift. We therefore predicted that a
gift giver’s thoughts would count—that is, would affect a receiv-
er’s gratitude and appreciation positively—when the receiver was
given a disliked gift by a close other, but would not count when the
receiver was given liked gifts from anyone or disliked gifts from
distant others.
This prediction highlights an important theoretical distinction
between the activation of a psychological process, such as activat-
ing the capacity to reason about the minds of others, and the
application of that process, such as using inferences about others’
mental states in judgment, evaluation, or choice. The distinction
between activation and application is fundamental in research on
the influence of stereotypes in social judgment (e.g., Gilbert &
Hixon, 1991; Kunda, 1999), the influence of stored associations in
behavior (Higgins, 1996), or the impact of systematic versus
heuristic thinking in decision making (Chen & Chaiken, 1999).
However, in research investigating mental state inferences—
typically referred to as using one’s theory of mind—the first stage
of activation is often overlooked. Instead, the bulk of research on
theory of mind investigates the subsequent application of this
capacity once people are implicitly or explicitly triggered to make
inferences about others’ minds, such as by being asked to predict
another person’s behavior (Birch & Bloom, 2007; Wimmer &
Perner, 1983) or to explain another person’s behavior (Ames et al.,
2004; Heider & Simmel, 1944). Even the widely cited Heider and
Simmel (1944) study that is commonly used (e.g., Guthrie, 1993)
to show how people attribute minds spontaneously to almost
anything, in this case to geometrical shapes, did so in a context
where people were asked to explain the behavior of these shapes.
In other research, mental state inferences can appear to be
activated automatically, but these results arise in paradigms where
participants are presented simple and unambiguous sentences in
which a person’s behavior or mind is the only thing that needs to
be evaluated (Van Overwalle, Van Duynslaeger, Coomans, &
Timmermans, 2012). When the person is the focus of attention in
a social setting, mental state inferences do indeed appear to be
relatively faster than other judgments made about a person (Malle
& Holbrook, 2012). But this does not mean that mental state
inferences are activated automatically. In a more complicated
social interaction where an object is likely to be the focus of
attention, such as a gift exchange, another person’s mental states
may not be considered at all.
The existing research strategies used to study mental state
inferences could therefore make them appear more spontaneous
than they actually are or even suggest that they represent a distinct
neural module. According to Scholl & Leslie (1999), for instance,
the “theory of mind mechanism is essentially a module which
spontaneously and post-perceptually attends to behaviors and in-
fers (i.e., computes) the mental states which contribute to them” (p.
147). And according to Frith & Frith (2012), “keeping track of
these mental states . . . occurs automatically and without the need
for awareness” (p. 299; see also Cohen & German, 2009; Sperber
& Wilson, 2002; Stone, Baron-Cohen, & Knight, 1998; Uleman,
Saribay, & Gonzalez, 2008).
We believe the experiments reported in this article provide a
theoretical test of the automaticity of mental state inferences. If
reasoning about other minds is spontaneous and automatic as some
researchers argue, then the amount of thought should influence a
receiver’s appreciation regardless of the gift’s objective quality. If,
however, reasoning about others’ mental states must be triggered
or activated (Apperly & Butterfill, 2009; Epley, Keysar, Van
Boven, & Gilovich, 2004; Epley, Morewedge, & Keysar, 2004;
Keysar & Barr, 2002; Zaki & Ochsner, 2011), then a gift giver’s
thoughts should count only when they are activated for gift receiv-
ers.
Mispredicted Impact: Gift Givers Fail to Anticipate
When Their Own Thoughts Count
If gift receivers consider a gift giver’s thoughts only when
triggered to do so, then gift givers are likely to have difficulty
correctly anticipating the impact of their own thoughts and inten-
tions. Gift givers, after all, have a very different perspective on
the exchange than do gift receivers. Gift givers do not experience
the automatic evaluation that comes from receiving a gift and thus
would not be triggered to use their thoughts to predict a receiver’s
feelings and evaluations in the same way as gift receivers.
Thoughts may indeed count for gift receivers, but not necessarily
in ways that givers will predict. The capacity to anticipate a
receiver’s feelings and evaluations is a critical aspect of gift
exchanges because maximizing the receiver’s happiness and sat-
isfaction is arguably the most common objective in gift exchanges
(Cheal, 1986, 1988). Any gap between a gift giver’s predictions
and a receiver’s actual evaluations will undermine the primary
goal of gift exchanges.
Misplaced Benefits: Thoughts Count for Gift Givers
by Increasing Relational Connection
The common wisdom underlying “it’s the thought that counts”
places the impact of a gift giver’s thoughts clearly within the
reactions of a gift receiver, but we predicted that the impact of a
gift giver’s thoughts should count most consistently on the person
actually having the thoughts, namely the gift giver. In particular,
we predicted that putting a lot of thought into a gift would
strengthen the social relationship by making the gift giver feel
socially closer to the gift receiver for two main reasons. First,
people typically spend more time and effort trying to please those
668 ZHANG AND EPLEY
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

they like than those they dislike. The amount of time and effort
spent trying to identify an ideal gift, even when experimentally
induced, may therefore be interpreted by gift givers as a sign of
liking and relationship closeness. Just as people can use their
behavior as a guide when assessing their own attitudes and pref-
erences (Bem, 1972), so too can people use the nature of their own
thoughts as a guide (Critcher & Gilovich, 2010). Second, spending
time and effort trying to identify an ideal gift is also likely to
require a considerable amount of perspective taking, trying to put
oneself in another’s shoes and imagine his or her evaluation.
Perspective taking tends to strengthen relational bonds (Galinsky,
Ku, & Wang, 2005), increasing perceived closeness and similarity
between people (Davis, Conklin, Smith, & Luce, 1996).
We predicted that a gift giver’s thoughts would not, however,
influence relational closeness for gift receivers simply because gift
receivers do not experience the effort of expending thoughts or the
process of perspective taking themselves. In this way the common
wisdom is somewhat misplaced. Putting careful thought and at-
tention into a gift benefits the giver, in different and potentially
more consistent ways than it benefits the receiver.
The Current Experiments
In everyday life, putting thought into a gift may be systemati-
cally confounded with choosing a better gift, making it appear that
a giver’s thoughts count when in fact it is the quality of the gift that
really counts. Gift givers may therefore think that “thoughts count”
in gift exchanges simply because it leads them to choose a better
gift, rather than producing additional gratitude and appreciation for
the gift receiver beyond the quality of the gift itself.
To test whether people truly believe that “thoughts count” in gift
exchanges beyond improving the quality of the gift itself, we
conducted an experiment with 44 visitors to the Museum of
Science and Industry in Chicago (the MSI) who agreed to partic-
ipate in exchange for a small prize. These participants predicted
how much appreciation and gratitude a gift giver would feel (on
separate scales ranging from 0 to 10) after receiving a birthday gift
from a friend in each of four different circumstances (order was
randomized) in a 2 (gift quality: liked vs. disliked) ⫻2 (thought:
thoughtful vs. thoughtless) within-participants design. Participants
were asked to imagine that someone bought a gift for a friend
thoughtfully (“putting a lot of careful thought into what the friend
would like and what the friend would dislike, choosing very
carefully from the large number of gifts that he or she could
purchase for the friend”) or thoughtlessly (“without putting any
careful thought into what the friend might like or dislike, and
simply choos[ing] a gift while at a store almost randomly from the
gifts that are available”). In each scenario, participants also learned
that when the friend opens the gift, it turns out that the friend either
“likes the gift very much” or “dislikes the gift very much.” These
scenarios allow a direct test of how much participants from this
population (that we sample from in Experiment 1–4) think their
thoughts count in gift exchanges relative to the quality of the gift
itself.
Participants predicted appreciation and gratitude was highly
correlated (r⫽.69), so we collapsed these measures into a single
positive evaluation composite. A 2 (gift quality: liked vs. dis-
liked) ⫻2 (thought: thoughtful vs. thoughtless) repeated-measures
analyses of variance (ANOVA) on positive evaluations revealed
significant main effects of similar magnitude for both the quality
of the gift, F(1, 43) ⫽138.22, p⬍.01, ␩
p
2
⫽.76, as well as the
amount of thought put into the gift, F(1, 43) ⫽104.27, p⬍.01,
␩
p
2
⫽.71. Not surprisingly, participants believed that a liked gift
would produce more appreciation and gratitude (M⫽7.51, SD ⫽
2.42) than a disliked gift (M⫽4.38, SD ⫽2.51). More important,
and consistent with the common wisdom that thought counts,
participants also believed that gifts chosen thoughtfully would
produce more appreciation and gratitude (M⫽7.44, SD ⫽2.37)
than gifts chosen thoughtlessly (M⫽4.45, SD ⫽2.63). The
interaction was nonsignificant, F(1, 43) ⫽0.31, p⫽.58. These
participants believed a gift giver’s thoughts would count for gift
receivers above and beyond simply choosing a better or worse gift.
We designed four experiments to test our three main hypotheses
about precisely when and for whom thoughts actually count in gift
exchanges, using procedures that separate perceptions of the qual-
ity of a gift and a gift giver’s thoughts in order to disentangle the
impact of each in both givers’ and receivers’ evaluations. We
utilized naturally occurring gift exchanges recalled from memory
in Experiments 1 and 2 and utilized experimentally manipulated
gift exchanges in Experiments 3 and 4. Experiment 1 tested our
hypothesis that a giver’s thoughts would count for receivers only
when they received a disliked gift, and our hypothesis that givers
would be unable to predict correctly when their thoughts count and
when they do not. Experiment 2 tested our prediction that thoughts
count in gift exchanges only when mental state inferences are
activated by leading one group of participants to evaluate a gift
giver’s thoughts explicitly before evaluating the exchange. Exper-
iment 3 tested a more focused prediction that thoughts only count
when receivers are given a bad gift from a friend or loved one, but
not from a stranger. Experiment 4 tested our prediction that
thoughts count for givers, but not for receivers, by increasing
relational closeness. Collectively, these experiments sharpen our
understanding of the role of mental state inferences in social
interactions, identifying when inferences about a person’s thoughts
and intentions influence evaluations and when they do not in an
important domain of everyday social life.
Experiment 1: Givers and Receivers
In Experiment 1, we surveyed a broad range of naturally occur-
ring gifts by asking gift givers and gift receivers to recall a liked
or disliked gift. Receivers reported how much appreciation and
gratitude they felt, and givers reported how much appreciation and
gratitude they believed the receivers felt. We expected that the
amount of thought gift givers expended would “count” for receiv-
ers—that is, would be correlated with appreciation and gratitude—
only when the gift was undesirable, but that givers would not
anticipate correctly when their thoughts counted and when they did
not.
Method
Ninety-nine visitors (48 female; M
age
⫽37.50 years) to the MSI
were randomly assigned to recall a time that they either gave or
received a gift. Givers recalled a gift that they believed the receiver
liked or disliked. Receivers recalled a gift that they actually liked
or disliked.
669
HOW THOUGHTS COUNT?
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

To measure how much thoughts “counted,” we asked gift re-
ceivers to report how much they appreciated the gift, how grateful
they felt, and how much they liked the gift giver after receiving the
gift on a scale ranging from 1 (not at all)to9(very much). Gift
givers answered the same three questions in terms of how they
believed the receiver felt. To measure “thoughts,” gift givers
reported how much thought they put into the gift and how much
they cared about whether the receiver would like the gift on the
same 1 to 9 scale. Gift receivers answered the same two questions
in terms of how much thought and care they believed givers
expended.
Finally, participants reported (or estimated) the gift’s cost. As a
manipulation check, receivers reported how much they liked the
gift and givers estimated how much the receiver liked the gift on
the same 1 to 9 scale.
Results and Discussion
All relevant means, standard deviations, and correlations within
experimental conditions are reported in Table 1.
Manipulation check. The recall manipulation was effective.
A 2 (role: giver vs. receiver) ⫻2 (gift quality: liked vs. disliked)
ANOVA on participants’ reported liking for the gift itself yielded
only a significant main effect for gift quality, F(1, 95) ⫽272.65,
p⬍.01, ␩
p
2
⫽.74. Participants reported liking the liked gift
significantly more (M⫽8.44, SD ⫽0.94) than the disliked gift
(M⫽3.28, SD ⫽2.02). No other effects approached significance.
Thoughts and evaluations. We averaged the three positive
evaluation measures (␣⫽.87) and the two thought measures (r⫽
.55) into two composites. A 2 (role: giver vs. receiver) ⫻2 (gift
quality: liked vs. disliked) ANOVA on participants’ positive eval-
uations yielded a significant main effect for gift quality, F(1, 95) ⫽
62.67, p⬍.01, ␩
p
2
⫽.40. Receivers who recalled a liked gift
reported more positive evaluations (M⫽8.48, SD ⫽1.09) than
those who recalled a disliked gift (M⫽5.39, SD ⫽1.77), F(1,
95) ⫽61.41, p⬍.01, ␩
p
2
⫽.39. Similarly, givers who recalled a
gift liked by the receiver predicted more positive evaluations from
receivers (M⫽7.81, SD ⫽1.24) than those who recalled a gift
disliked by the receiver (M⫽6.45, SD ⫽1.41), F(1, 95) ⫽11.52,
p⬍.01, ␩
p
2
⫽.11. This main effect was qualified by a Role ⫻Gift
Quality interaction, F(1, 95) ⫽9.49, p⬍.01, ␩
p
2
⫽.09, indicating
that the difference in evaluations was larger for receivers than for
givers. This interaction was not predicted, does not bear directly on
our hypotheses about how much thoughts count in gift exchanges,
does not replicate in an identical condition in Experiment 2, and
does not replicate in conceptually similar conditions in Experiment
3 and 4. We will therefore not discuss it further.
Our main prediction was that ratings of gift givers’ thoughts
would be significantly correlated with gift receivers’ positive
evaluations only when recalling a disliked gift. As predicted, the
amount of thought that receivers believed their givers expended
was significantly correlated with their positive evaluation of the
exchange when recalling a disliked gift, r(23) ⫽.80, p⬍.01, but
not when recalling a liked gift, r(23) ⫽⫺.14, p⫽.50, z⫽4.11,
p⬍.01. Controlling for the perceived cost of the gift did not alter
the results meaningfully (rs⫽.79 & ⫺.18, disliked vs. liked,
respectively, p⬍.01 & p⫽.39, z⫽4.16, p⬍.01).
We also predicted that gift givers would fail to anticipate cor-
rectly when their thoughts counted in receivers’ evaluations and
when they did not. Consistent with this prediction, gift givers
showed a very different pattern of expected evaluations than gift
receivers. In particular, gift givers believed that their thoughts
would count—that is, would increase a receivers’ positive evalu-
ations—when giving a liked gift, r(25) ⫽.47, p⫽.01, and showed
a similar but nonsignificant pattern when giving a disliked gift,
r(20) ⫽.28, p⫽.20, z⬍1. Controlling for the cost of the gift did
not alter the results meaningfully (rs⫽.41 & .13, liked vs.
disliked, respectively, ps⫽.04 & .56, z⬍1.). Gift givers expected
their thoughts to count when giving a good gift but to count
somewhat less when giving a bad gift. For receivers, it was
precisely the opposite.
One alternative interpretation is that a gift giver’s thoughts did
not count when receiving a good gift because of a ceiling effect or
a restricted range. That is, liked gifts were evaluated so positively
that there was not enough variability in our measures for a positive
correlation to emerge, whereas disliked gifts were evaluated more
negatively with enough sensitivity and variability around the mean
for a significant correlation to emerge. This alternative posits that
the correlations in these two distributions do not actually differ
significantly from each other, but that the correlation in the liked
gift condition is suppressed because of censored data at the ceiling
of the positive evaluation or thoughts measures.
Indeed, we have markedly more evaluations at ceiling among
gift receivers in the liked gift condition (72.0%) than in the
disliked gift condition (0.0%), and more than among gift givers in
both the liked gift (44.4%) and disliked gift (13.6%) conditions.
This result alone is not inconsistent with our account. Our predic-
tion is that gift receivers do not consider a gift giver’s thoughts
when receiving a good gift. This could produce positive evalua-
tions at the ceiling regardless of gift receivers’ presumed thoughts.
However, the result would be entirely inconsistent with our ac-
count if these ceiling effects are masking a strong positive corre-
Table 1
Average Positive Evaluations, Thoughts, and the Correlation Between Evaluations and Thoughts for Receivers and Givers Evaluating
Liked Versus Disliked Gifts (Experiment 1)
Group
Liked gift Disliked gift
Positive evaluation Thoughts rPositive evaluation Thoughts r
Receivers 8.48 (1.09) 7.50 (1.84) ⫺.14 5.39 (1.77) 5.50 (2.29) .80
*
Givers 7.81 (1.24) 7.52 (1.60) .47
*
6.45 (1.41) 6.14 (1.75) .28
Note. Values in the parentheses are standard deviations. Asterisks indicate correlations significantly larger than zero. p⬍.05.
670 ZHANG AND EPLEY
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

lation between thoughts and evaluations that could emerge with
more sensitive measures in the liked gift condition.
We assessed the validity of this alternative interpretation in
three different ways. We address each of these ways concisely here
but provide a full description of all of our additional analysis and
the scatterplots for all 12 conditions in both Experiments 1 and 2
in the supplemental online materials.
First, we assessed whether the range restrictions in our data are
so severe for liked gifts among receivers that a positive correlation
could not even theoretically emerge in that condition. To do so, we
calculated the maximum possible correlation between positive
evaluations and thoughts by breaking the dependence of the two
measures within participants and then rank ordering the observa-
tions on each measure to identify the maximum positive and
negative correlations. Given the data, a large positive correlation
was possible in both the liked gift (r⫽.95, p⬍.01) and disliked
gift (r⫽.95, p⬍.01) conditions among receivers, suggesting that
it is possible to achieve a highly significant correlation for liked
gifts, even when a large number of data points are at ceiling. The
maximum negative correlation for gift receivers was smaller in the
liked gift condition (r⫽⫺.41, p⬍.05) than in the disliked gift
condition (r⫽⫺.95, p⬍.01). At the very least, a meaningful
positive correlation was possible for gift receivers recalling liked
gifts given the range of evaluations we observed.
Second, the alternative interpretation based on a ceiling effect
makes one prediction that our account does not. In particular, it
predicts that if one was able to raise the ceiling on either the
positive evaluation or thoughts measures, then a positive correla-
tion would emerge. A Tobit regression is designed for exactly this
situation in which some data points are censored by ceiling (or
floor) effects by treating data points at the ceiling as the maximum
value or greater. This analysis conceptually lifts the ceiling on a
potentially truncated distribution. Conducting Tobit regressions on
our data did not support the existence of a suppressed positive
correlation for gift receivers evaluating a liked gift. As before, the
amount of thought that receivers believed their givers expended
was significantly correlated with their positive evaluation of the
exchange when recalling a disliked gift (B⫽.55, p⬍.001), but
there was still no significant relationship between thoughts and
positive evaluation when receivers recalled a liked gift (B⫽⫺.38,
p⫽.41). The directionally negative relationship for liked gifts
among gift receivers further suggests that no positive correlation is
suppressed by a ceiling effect. For gift givers, Tobit regressions
again suggest patterns very different from gift receivers. The
amount of thought givers reported expending was not significantly
correlated with their predicted positive evaluations when recalling
a disliked gift (B⫽.23, p⫽.17), but was significantly correlated
when recalling a liked gift (B⫽.43, p⫽.02).
Another way to increase the ceiling on our measures would be
to log-transform the data to normalize a skewed distribution.
Because our data are negatively skewed, with a relatively large
percentage of receivers rating either thoughts or positive evalua-
tions at ceiling, we conducted the log transformation on the two
measures by first reflecting the data, adding 10 to bring the lowest
possible value to 1, taking the log, reflecting back to the original
order of the measure, and adding another constant to ensure the
transformed data were positive (that is, Measure
transformed
⫽
⫺log(⫺Measure
original
⫹10) ⫹constant). Normalizing the data in
this way did not meaningfully alter the correlation between
thoughts and positive evaluation for gift receivers in the liked gift
condition, r(23) ⫽⫺.05, p⫽.80.
Third, the alternative interpretation based on a ceiling effect
makes a second prediction that our account does not. In particular,
it predicts that the distribution for disliked gifts would look the
same as it does for liked gift if it, too, was subject to a ceiling
effect. We tested this possibility by lowering the ceiling on the
positive evaluation and thoughts measures for disliked gifts to see
if artificially creating a ceiling effect in these conditions produces
results similar to those in the liked gift conditions. We truncated
the distribution for gift receivers in the disliked condition in many
different ways, from the most mild truncation that matches the
number of data points at the most extreme points of the distribution
in the liked gift condition, to a more severe truncation that matches
the percentage of observations at ceiling in the liked gift condition,
to even more severe truncations than we actually observe in the
liked gift condition. At no point does the correlation we observe in
the disliked condition resemble the correlation of ⫺.14 that we
observe in the liked gift condition. The lowest correlation we
observe is .64, with a very severe truncation that censors 80% of
the observations on the positive evaluation measure and 76% of
the evaluations on the thoughts measure (far more censoring than
we actually observe among gift receivers in the liked gift condi-
tion). All other truncations yield stronger positive correlations.
None of these results suggest a suppressed correlation resulting
from ceiling effects or range restrictions.
Experiment 2: Activating Thoughts
Collectively, we believe the results from Experiment 1 suggest
that thoughts “count” for gift receivers only when they are trig-
gered to think of a gift giver’s thoughts. However, the results of
Experiment 1 are also consistent with an alternative interpretation
based on the application of mental state inferences rather than on
their activation (e.g., Gilbert & Hixon, 1991). It could be that gift
receivers actively considered a gift giver’s thoughts when receiv-
ing a good as well as a bad gift in Experiment 1, but only applied
those inferences to their evaluations when considering a bad gift.
Simply measuring the correlation between inferences about a gift
giver’s thoughts and evaluations of the exchange cannot distin-
guish between the activation of mental state inferences and the
application of those inferences in judgment.
Unlike the application account, our activation account suggests
that a gift giver’s presumed thoughts do not influence gift receiv-
ers’ evaluations because they are less accessible than the objective
quality of the gift itself, not because gift receivers think a giver’s
thoughts are irrelevant to their evaluations. Indeed, the pretest data
we reported in the introduction suggests that people, on average,
have a strong theory that gifts chosen thoughtfully will be appre-
ciated above and beyond the objective quality of the gift itself.
These data suggest that a gift giver’s thoughtfulness is indeed seen
as applicable to a gift receiver’s appreciation and gratitude.
To test whether or not the results from Experiment 1 stemmed
from the activation of mental state inferences, we manipulated
whether or not a gift giver’s thoughts were activated when receiv-
ers evaluated the gift exchange. We did so in Experiment 2 by
asking one group of gift receivers to evaluate a gift giver’s
thoughts before reporting their appreciation and gratitude and
another group to do so after reporting their appreciation and
671
HOW THOUGHTS COUNT?
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

gratitude. If a gift giver’s thoughts count only when inferences
about them are activated at the time of evaluation, then gift
receivers’ thoughts should count for both liked as well as disliked
gifts when people are led to think of them before reporting appre-
ciation and gratitude, but should only count for disliked gifts when
receivers are led to think of them after reporting appreciation and
gratitude. If, however, the results of Experiment 1 were produced
by differences in the application of a gift receiver’s thoughts in
evaluations of appreciation and gratitude, then a gift giver’s
thoughts should count only for disliked gifts regardless of whether
the gift receiver was asked to think about a gift giver’s thoughts
before or after evaluating appreciation and gratitude.
Method
A total of 161 MSI visitors (91 female, M
age
⫽39.5 years)
volunteered to participate in the experiment in exchange for a
small prize. Experiment 2 was identical to Experiment 1, except
that we manipulated the order of the questions about a gift giver’s
thoughts and the questions about positive evaluations of the ex-
change. Participants were randomly assigned to the role of giver or
receiver and then recalled either a liked gift or a disliked gift. After
that, participants randomly assigned to the thought-first condition
evaluated the thought items before the positive evaluation items,
whereas those in the thought-last condition answered the items in
the opposite order (as in Experiment 1).
Results and Discussion
All relevant means, standard deviations, and correlations within
experimental conditions are reported in Table 2.
Manipulation checks. As in Experiment 1, the recall manip-
ulation was effective. A 2 (role: giver vs. receiver) ⫻2 (gift
quality: liked vs. disliked) ⫻2 (order: thought first vs. thought
last) ANOVA on participants’ reported liking for the gift itself
yielded only a significant main effect for gift quality, F(1, 153) ⫽
311.11, p⬍.01, ␩
p
2
⫽.67. Participants reported liking the liked
gift significantly more (M⫽8.44, SD ⫽1.20) than the disliked
gift (M⫽3.73, SD ⫽2.26). No other effects approached signif-
icance.
Thoughts and evaluations. Because they were highly corre-
lated, we again averaged the three appreciation measures (␣⫽.90)
and the two thought measures (r⫽.66) into two composites. A 2
(role: giver vs. receiver) ⫻2 (gift quality: liked vs. disliked) ⫻2
(order: thought first vs. thought last) ANOVA on participants’
positive evaluations yielded a significant main effect for gift
quality, F(1, 153) ⫽184.88, p⬍.01, ␩
p
2
⫽.55, and a marginally
significant main effect for role, F(1, 153) ⫽3.49, p⫽.06, ␩
p
2
⫽
.02. Givers expected more positive evaluations when recalling a
liked gift (M⫽7.89, SD ⫽1.49) than when recalling a disliked
gift (M⫽4.72, SD ⫽1.76), F(1, 153) ⫽84.11, p⬍.01, ␩
p
2
⫽.35,
and gift receivers provided a similar pattern (M
liked
⫽8.47, SD ⫽
0.80 vs. M
disliked
⫽5.03, SD ⫽1.84), F(1, 153) ⫽101.34, p⬍
.01, ␩
p
2
⫽.40, but were slightly more positive overall than gift
givers. No other effects approached statistical significance.
Most important again were the correlations within conditions
between the thought and positive evaluation measures. Results
from the thought-last condition replicated the main findings of
Experiment 1. Specifically, the amount of thought that receivers
believed their givers expended was significantly correlated with
their positive evaluations (appreciation, gratitude, and liking of the
gift giver) when recalling a disliked gift, r(19) ⫽.59, p⬍.01, but
not when recalling a liked gift, r(17) ⫽.01, p⫽.95, z⫽1.95, p⫽
.05, suggesting that a giver’s thoughts only counted by increasing
a receiver’s appreciation for bad gifts but not for good gifts.
Controlling for the perceived cost of the gift did not alter the
results meaningfully (rs⫽.57 & ⫺.004, disliked vs. liked, respec-
tively, p⬍.01 & p⫽.99, z⫽1.90, p⫽.06).
For gift givers, however, the evaluation of their own thoughts
was not correlated with their predicted positive evaluations from
receivers either when recalling a disliked gift, r(19) ⫽⫺.14, p⫽
.54, or a liked gift, r(18) ⫽.29, p⫽.24, z⫽1.30, ns. Controlling
for the cost of the gift did not meaningfully alter the results (rs⫽
⫺.14 & .25, disliked vs. liked, respectively, ps⫽.57 & .31, z⫽
1.17, ns). Again, thoughts only counted for those who received bad
gifts, an effect not anticipated by gift givers.
A very different pattern of correlations emerged among partic-
ipants in the thought-first condition. In this condition, the amount
of thought that receivers believed their givers expended was sig-
nificantly correlated with their positive evaluation for both disliked
gifts, r(20) ⫽.61, p⬍.01, and liked gifts, r(17) ⫽.62, p⬍.01,
z⬍1, suggesting that a giver’s thoughts counted by increasing a
receiver’s appreciation for both disliked and liked gifts. Control-
ling for the cost of the gift did not meaningfully alter the results
(rs⫽.70 & .63, disliked vs. liked, respectively, ps⬍.01, z⬍1).
Interestingly, a roughly similar pattern emerged for gift givers as
well: the amount of thought givers expended was positively cor-
Table 2
Average Positive Evaluations, Thoughts, and the Correlation Between Evaluations and Thoughts for Receivers and Givers Evaluating
Liked Versus Disliked Gifts Among Those Who Considered a Gift Giver’s Thoughts First Versus Last (Experiment 2)
Condition/group
Liked gift Disliked gift
Positive evaluation Thoughts rPositive evaluation Thoughts r
Thoughts first
Receivers 8.32 (0.98) 7.58 (1.38) .62
ⴱ
4.83 (1.88) 4.95 (2.13) .61
ⴱ
Givers 7.84 (1.86) 7.74 (1.28) .49
ⴱ
4.32 (1.39) 6.18 (1.92) .35
Thoughts last
Receivers 8.62 (0.58) 7.68 (1.08) .01 5.24 (1.82) 5.07 (2.45) .60
ⴱ
Givers 7.93 (1.07) 7.78 (1.04) .29 5.10 (2.00) 6.93 (1.76) ⫺.14
Note. Values in the parentheses are standard deviations. Asterisks indicate correlations significantly larger than zero. p⬍.05.
672 ZHANG AND EPLEY
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

related with their predicted evaluations for both disliked gifts,
r(17) ⫽.35, p⫽.14, and for liked gifts, r(17) ⫽.49, p⫽.03.
Controlling for the cost of the gift did not meaningfully alter these
results (rs⫽.28 & .52, disliked vs. liked, respectively, ps⫽.26
& .03, z⬍1).
As in Experiment 1, we further assessed whether these patterns
of correlations could be produced by a ceiling effect or range
restriction. Please see the supplemental online materials for a
complete description of these analyses. First, the maximum pos-
sible positive correlations given the data we observed were equally
high in all experimental conditions (all maximum positive rs⬎.9).
Second, conducting Tobit analyses on our data did not alter the
results in any meaningful way. Third, normalizing the data by a log
transformation did not meaningfully alter the observed correla-
tions. Fourth, creating a ceiling effect in the distribution for gift
receivers recalling a disliked gift in the thought-last condition did
not reduce the correlation to the level we observed among gift
receivers recalling a liked gift in the thought-last condition. Most
important, however, the significantly different pattern of correla-
tions among gift receivers evaluating liked gifts in the thought-first
versus thought-last condition was inconsistent with a ceiling effect
explanation. When the process of reasoning about a giver’s
thoughts was activated by triggering gift receivers to think of a gift
giver’s thoughts, their positive evaluations were correlated with
the giver’s presumed thoughts. When the process was not explic-
itly activated, positive evaluations and presumed thoughts were
unrelated. The range of responses in these two conditions on the
thought measure was nearly identical, but the relation of responses
to the positive evaluation of the exchange differed markedly.
We believe these results are interesting for two reasons. First,
they are consistent with our theory that thoughts count in gift
receivers’ evaluations only when mental state inferences are acti-
vated. In everyday life, we believe a common trigger for this
process is the automatic evaluation of the gift itself. If receivers
like a gift, then a giver’s thoughts are not considered. But if
receivers dislike the gift, then additional cognitive processing of
the exchange is activated in order to explain the behavior, leading
people to consider the giver’s thoughts. Experiment 2 replicated
the basic finding from Experiment 1, but also demonstrated that
evaluating a gift giver’s thoughts explicitly before reporting ap-
preciation and gratitude led gift receivers to take into account a
giver’s thoughts. These results suggest that a gift giver’s thoughts
do not count in a gift receiver’s evaluations because the receiver
fails to consider them at the time of judgment, rather than because
the gift receiver thinks that the giver’s thoughts are irrelevant or
because they are unable to consider them.
Second, these results support our hypothesis that giver and
receiver evaluations may naturally diverge because of different
perspectives on the gift exchange. When both gift givers and gift
receivers were focused on the gift givers’ thoughts before report-
ing or predicting positive evaluations of the exchange, gift givers’
thoughts counted at least somewhat for givers and receivers for
both liked as well as disliked gifts. When receivers were not asked
to consider a gift giver’s thoughts before evaluating the exchange,
perspectives between givers and receivers diverged. “It’s the
thought that counts” for gift receivers only when those thoughts
are activated at the time of evaluation. Otherwise, it is the quality
of the gift that counts first and foremost.
Although these results are consistent with our hypotheses, one
caveat deserves mentioning. In Experiment 2, we manipulated
whether a gift giver’s thoughts were accessible at the time of
judgment but measured how inferences about those thoughts were
related to evaluations of gratitude and appreciation. We did not
measure the activation of mental state inferences directly. It is
theoretically possible that gift receivers did consider how much
thought a gift giver expended in all conditions, but that those
activated inferences did not influence appreciation and gratitude
for liked gifts in the thought-last condition. We think this is
implausible because it requires an existing theory among gift
receivers that a gift giver’s thoughts are relevant only for relatively
bad gifts. The pretest described in the introduction showed no
evidence for such a belief whatsoever. Instead, thoughtful gifts
were expected to be appreciated more than thoughtless gifts in all
cases. We therefore believe that Experiment 2 is most parsimoni-
ously explained by differences in the activation of mental state
inference, but note that a completely conclusive test would require
both manipulating and measuring the activation of these inferences
directly.
The strength of Experiments 1 and 2 is that they utilized eval-
uations of naturally occurring gifts, thereby providing a reasonable
degree of ecological validity. The weakness of Experiments 1 and
2 is that they both relied on memory for gift exchanges rather than
on the immediate experience of giving or receiving a gift. It is
possible that gifts recalled from memory at a distance are evalu-
ated differently than gifts experienced directly in an online ex-
change. In particular, it is possible that the relative accessibility of
a gift giver’s thoughts compared with the objective quality of the
gift itself may vary, with a gift giver’s thoughts being less acces-
sible to both givers and receivers in memory than they might be
immediately. The gifts recalled from memory in Experiments 1
and 2 also varied between the liked and disliked conditions,
making it possible that some variable other than the objective
quality of the gift (whether it was liked or disliked) is responsible
for the differences we observed. It is therefore critical to conduct
experiments involving real gift exchanges in which evaluations are
measured immediately instead of recalled from memory and in
which the actual gifts exchanged do not vary systematically by
condition. The final two experiments utilized an experimental
procedure designed to meet these goals.
Experiment 3: Actual Gift Exchange
In addition to providing an experimental design that allowed for
tighter control and stronger causal inferences, Experiment 3 also
provided a further test of our prediction that gift givers’ thoughts
would count only when mental state reasoning was activated by
manipulating whether the gift giver was a friend or stranger. In
Experiments 1 and 2, gifts were exchanged between friends or
family members, as occurs routinely in everyday life. We pre-
sumed people would have relatively strong expectations about
getting good gifts from friends but have relatively weaker expec-
tations about getting good gifts from strangers. As a result, we
predicted that a bad gift from a friend would activate the process
of considering the gift giver’s thoughts but that a bad gift from a
stranger would not. This friend versus stranger manipulation
would provide not only another test of our suggested psychological
673
HOW THOUGHTS COUNT?
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

mechanism, but also test another important boundary condition for
when thoughts count in gift exchanges.
To test our critical assumption that gift receivers expect better
gifts from friends than from strangers, we recruited 44 visitors to
the MSI in Chicago to participate in a short experiment. Each
participant named a close friend and then reported the extent to
which they expected to like a birthday gift from this friend on a
scale ranging from ⫺5(strongly expect to dislike)to5(strongly
expect to like) and also how likely this friend would be able to
choose a gift that they really liked on a scale ranging from 0 (not
likely at all)to10(very likely). Participants did exactly the same
for a distant acquaintance, with the order of consideration coun-
terbalanced.
Consistent with our assumption, participants expected to like the
gift from a close friend (M⫽3.31, SD ⫽1.62), but had no
consistent expectation for a gift from a distant acquaintance (M⫽
0.33, SD ⫽1.88), t(43) ⫽8.42, p⬍.01, r⫽.79. Participants also
predicted that a close friend would be better able to choose a gift
that they would like (M⫽8.02, SD ⫽1.82) than would a distant
acquaintance (M⫽4.66, SD ⫽2.29), t(43) ⫽10.04, p⬍.01, r⫽
.84. If mental state reasoning is activated by unexpected events,
then it should be activated when a friend gives a disliked gift but
not when a stranger does so. And if mental state inferences count
only when activated in judgment, then a gift giver’s thoughts
should count only when a close friend gives a disliked gift.
We tested this prediction by asking MSI visitors to exchange
gifts either with an acquaintance or with a stranger. Gift givers
selected a gift from a set of five options after seeing an example
gift that was either very desirable or very undesirable to make the
actual gifts seem relatively bad or good by comparison. We ex-
pected, consistent with evidence on the importance of comparison
standards in judgment (e.g., Mussweiler, 2003), that our partici-
pants would like the actual gifts more after considering the unde-
sirable example gift than after considering the desirable example
gift. In this way, the very same gift could be considered desirable
for some but undesirable for others.
To manipulate the amount of thought expended on the gift, we
asked gift givers in the thoughtful condition to choose a gift
carefully and with a lot of thought, but asked gift givers in the
thoughtless condition to choose randomly without any thought at
all. Receivers were told how givers were instructed to choose the
gift in detail, mirroring the gift giver’s instructions precisely.
To investigate whether this particular paradigm leads people to
believe a giver’s thoughts count, we conducted another pilot ex-
periment with 47 MSI visitors. After having Experiment 3’s pro-
cedure explained in detail and viewing the good or bad example
gift, these visitors acted as observers and predicted the receivers’
appreciation and gratitude after receiving a thoughtful versus
thoughtless gift from a friend versus stranger (on 0 to 10 scales,
averaged into a composite, r⫽72). As shown in Figure 1, a 2 (gift:
good vs. bad) ⫻2 (thought: yes vs. no) ⫻2 (relationship: ac-
quaintance vs. stranger) mixed-model ANOVA on positive eval-
uation revealed a main effect of thought, F(1, 45) ⫽30.69, p⬍
.01, ␩
p
2
⫽.41. Specifically, these observers believed that a gift
giver’s thoughts would count, increasing receivers’ appreciation
and gratitude regardless of whether the gift was good or bad by
comparison or whether the gift was coming from an acquaintance
or a stranger (M
thoughtful
⫽8.28, SD ⫽1.96 vs. M
thoughtless
⫽6.76,
SD ⫽2.31). The observers also believed that gifts from an ac-
quaintance would be appreciated more than gifts from a stranger,
(M
acquaintance
⫽7.96, SD ⫽2.32 vs. M
stranger
⫽7.08, SD ⫽2.14),
F(1, 45) ⫽8.29, p⬍.01, ␩
p
2
⫽.16, but this main effect did not
interact with the gift givers’ thoughts, F(1, 45) ⫽2.89, ns. The
conventional wisdom—“It’s the thought that counts”—appears
alive and well in this gift exchange paradigm.
Method
MSI visitors (N⫽336, 171 female, M
age
⫽33.3 years) were
first randomly assigned to participate with either an acquaintance
(a friend or family member attending the museum with them) or a
stranger (also attending the museum, but with another group of
friends or family members). Participants were then randomly as-
signed within the pair to be either a gift giver or gift receiver and
then led to separate rooms within the museum.
To manipulate whether a gift seemed relatively good or bad, we
first showed both givers and receivers either a very desirable
example gift (a “Newton’s Cradle”) or a very undesirable gift (a
small wooden ruler). These gifts were rated the most and least
desirable, respectively, out of eight gifts in a pretest we conducted
with 60 MSI visitors. The experimenter then explained that this
gift was merely an example of the kinds of gifts we used in the
experiment, but that it was not being used on that particular day.
Gift givers were then shown a set of five gifts to choose from (a
tote bag, a deck of cards, a big pen, a MSI keychain, and a MSI
magnet) of roughly equivalent desirability (based on the pretest of
60 MSI visitors). We expected that these gifts would seem rela-
tively bad compared with the desirable example gift, but relatively
good compared with the undesirable example gift.
Gift givers were then instructed to choose a gift for the receiver.
Givers in the thoughtful condition were asked to think very care-
fully about their choice, to put a lot of thought into choosing what
the receiver would like most. Givers in the thoughtless condition
were told that they should not think hard about their choice and
that it was fine to even choose randomly. Receivers were told
Figure 1. Observers’ predicted positive evaluations (appreciation and
gratitude, combined) when givers chose a good gift or a bad gift either
thoughtfully or thoughtlessly for an acquaintance or a stranger (pretest to
Experiment 3). Asterisks indicate a statistically significant simple effect
(p⬍.05) between thoughtful and thoughtless conditions. Error bars report
standard deviations.
674 ZHANG AND EPLEY
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

about the giver’s choice instructions in detail, using the same
instructions given to the givers.
After the experimenter delivered the chosen gift to the receiver,
givers reported how appreciative and grateful the receiver would
feel, whereas receivers reported how appreciative and grateful they
actually felt. Both groups used scales ranging from 0 (not at all)to
10 (very much). As a manipulation check, givers reported how
much thought they expended and how much they cared whether
the receiver would like the gift on the same 0 to 10 scale. Gift
receivers answered the same two questions in terms of how much
thought and care they believed givers expended. Receivers also
reported how much they liked the actual gift and how much they
liked the example gift, whereas givers predicted how much receiv-
ers would like the actual gift and the example gift. Finally, both
givers and receivers estimated the cost of the actual gift.
Results and Discussion
Manipulation check. Both givers and receivers liked the
good example gift significantly better than the bad example gift. A
complete 2 (role: giver vs. receiver) ⫻2 (thought: yes vs. no) ⫻
2 (gift quality: good vs. bad) ⫻2 (relationship: acquaintance vs.
stranger) mixed-model ANOVA with role treated as a repeated
measure returned a significant main effect for example gift, F(1,
160) ⫽125.23, p⬍.01, ␩
p
2
⫽.44. Specifically, givers liked the
Newton’s Cradle (M⫽6.15, SD ⫽1.82) more than the ruler (M⫽
2.89, SD ⫽1.59), F(1, 160) ⫽154.34, p⬍.01, ␩
p
2
⫽.49.
Receivers also liked the Newton’s Cradle (M⫽6.31, SD ⫽2.07)
better than the ruler (M⫽4.18, SD ⫽2.56), F(1, 160) ⫽34.41,
p⬍.01, ␩
p
2
⫽.18. This main effect was qualified by a Gift
Quality ⫻Role interaction, F(1, 160) ⫽7.41, p⬍.01, ␩
p
2
⫽.04,
indicating that the difference in example gift evaluations was
larger for givers than for receivers. No other effects approached
significance.
A similar ANOVA on liking of the actual gift also produced a
significant main effect for gift, F(1, 160) ⫽15.35, p⬍.01, with
the actual gift rated more favorably in the good gift condition (i.e.,
compared with the negatively rated ruler) than in the bad gift
condition (i.e., compared with the positively rated Newton’s Cra-
dle). This result, however, was qualified by a Gift Quality ⫻Role
interaction, F(1, 160) ⫽8.81, p⬍.01. This indicates that receivers
liked the actual gift significantly more in the good gift condition
(M⫽7.10, SD ⫽1.95) than in the bad gift condition (M⫽5.73,
SD ⫽2.07), F(1, 160) ⫽19.38, p⬍.01, but that the difference in
predicted liking by givers was nonsignificant between the good
gift condition (M⫽6.13, SD ⫽1.54) and the bad gift condition
(M⫽5.87, SD ⫽1.59), F(1, 160) ⫽1.14, p⫽.28. Gift receivers’
liking of the actual gift was more sensitive to the comparison gift
than were gift givers’ predictions of the receivers’ liking.
The thought manipulation was clearly effective. We averaged
the two thought measures into a composite because they were
highly correlated (r⫽.59). A 2 (role: giver vs. receiver) ⫻2
(thought: yes vs. no) ⫻2 (gift quality: good vs. bad) ⫻2 (rela-
tionship: acquaintance vs. stranger) mixed-model ANOVA re-
turned a significant main effect for thought, F(1, 160) ⫽22.55,
p⬍.01, ␩
p
2
⫽.12. As intended, givers in the thoughtful condition
reported that they invested more thought (M⫽6.28, SD ⫽1.86)
than did givers in the thoughtless condition (M⫽5.14, SD ⫽
2.39), F(1, 160) ⫽19.52, p⬍.01, ␩
p
2
⫽.11. Also as intended,
receivers were equally aware of this difference (F
interaction
⬍1).
Receivers in the thoughtful condition believed that givers ex-
pended more thought (M⫽5.99, SD ⫽2.04) than did receivers
in the thoughtless condition (M⫽4.95, SD ⫽2.29), F(1,
160) ⫽8.21, p⬍.01, ␩
p
2
⫽.05.
Positive evaluations. We averaged ratings of appreciation
and gratefulness into a single positive evaluation composite (r⫽
.68). As predicted, gift exchanges between acquaintances repli-
cated the main finding from the prior experiments. A 2 (role: giver
vs. receiver) ⫻2 (thought: yes vs. no) ⫻2 (gift quality: good vs.
bad) mixed-model ANOVA on positive evaluations yielded a main
effect of role, F(1, 79) ⫽13.71, p⬍.01, ␩
p
2
⫽.15, and a main
effect of thought, F(1, 79) ⫽4.00, p⬍.05, ␩
p
2
⫽.05, qualified by
a three-way interaction, F(1, 79) ⫽5.08, p⬍.05, ␩
p
2
⫽.05. As
shown in Figure 2, the givers’ thoughts had no influence on
receivers’ evaluations for a relatively good gift (M
thoughtful
⫽7.07,
SD ⫽1.38 vs. M
thoughtless
⫽7.47, SD ⫽1.42), F⬍1. However,
the givers’ thoughts influenced receivers’ appreciation and grati-
tude when receivers were given a relatively bad gift (M
thoughtful
⫽
7.38, SD ⫽1.03 vs. M
thoughtless
⫽5.74, SD ⫽2.11), F(1, 79) ⫽
11.99, p⬍.01, ␩
p
2
⫽.13, F
interaction
(1, 79) ⫽9.19, p⬍.01,
Figure 2. Expected and actual positive evaluations (appreciation and
gratitude, combined) when givers chose a good gift or a bad gift either
thoughtfully or thoughtlessly for an acquaintance (top panel) or a stranger
(bottom panel; Experiment 3). The asterisk indicates a statistically signif-
icant simple effect (p⬍.05) between the thoughtful and thoughtless
conditions. Error bars report standard deviations.
675
HOW THOUGHTS COUNT?
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

␩
p
2
⫽.10. As before, gift givers’ thoughts only counted for receiv-
ers when they received a relatively bad gift.
Among givers, in contrast, no significant main effects or inter-
actions emerged, Fs ⬍1.5, ns. As in the prior experiments, givers
did not anticipate correctly when their thoughts would influence
receivers’ appreciation and gratitude. In addition, a main effect of
role indicates that givers systematically underestimated receivers’
positive evaluations (M
giver
⫽6.11, SD ⫽1.54 vs M
receiver
⫽6.90,
SD ⫽1.66), F(1, 79) ⫽13.71, p⬍.01, ␩
p
2
⫽.15, replicating prior
research (Flynn & Brockner, 2003). This misprediction was espe-
cially large for good gifts chosen thoughtlessly (M
giver
⫽5.82,
SD ⫽1.83 vs. M
receiver
⫽7.47, SD ⫽1.42), F(1, 20) ⫽7.33, p⫽
.01, ␩
p
2
⫽.27.
When participants received a gift from a stranger, in contrast,
only the quality of the gift counted. Participants reported or pre-
dicted significantly more appreciation and gratitude when receiv-
ing a relatively good gift (M⫽6.53, SD ⫽1.62) than when
receiving a relatively bad gift (M⫽5.90, SD ⫽1.75), F(1, 81) ⫽
6.69, p⫽.01, ␩
p
2
⫽.08. This main effect was qualified by a
marginally significant Role ⫻Gift interaction, F(1, 81) ⫽3.57,
p⫽.06, ␩
p
2
⫽.04, indicating that the quality of the gift had a
somewhat larger effect on receivers’ evaluations (M
good
⫽7.33,
SD ⫽1.59 vs. M
bad
⫽6.24, SD ⫽2.09), F(1, 81) ⫽7.16, p⫽
.009, ␩
p
2
⫽.08, than it did on givers’ predicted evaluations
(M
good
⫽5.73, SD ⫽1.20 vs. M
bad
⫽5.56, SD ⫽1.27), F⬍1,
ns. Once again, the receivers also reported more gratitude and
appreciation (M⫽6.78, SD ⫽1.93) than givers expected (M⫽
5.64, SD ⫽1.23), F(1, 81) ⫽22.16, p⬍.01, ␩
p
2
⫽.21. No other
significant main effects or interactions emerged for either givers or
receivers. Most relevant for our hypotheses, a gift giver’s thoughts
did not count for gifts exchanged between strangers.
As in Experiments 1 and 2, putting careful thought into a gift
influenced receivers’ evaluations only when we predicted that
receivers would be triggered to activate the psychological process
of considering a giver’s thoughts, namely, when a friend gave
them a relatively bad gift. As before, givers could not predict this
correctly. The common wisdom that “it’s the thought that counts”
expressed by the observers in the pilot test for this experiment was
both miscalibrated and mispredicted. A gift giver’s thoughts did
not count in any measureable way when participants received a
good gift from anyone or a relatively bad gift from a stranger.
Experiment 4: Thoughts Count for Gift Givers
Gift receivers need to be triggered to think about a gift giver’s
thoughts in order for those thoughts to count in receivers’ evalu-
ations following the exchange. Gift givers, however, experience
their thoughtfulness directly and may therefore be influenced in
unique ways by the thought they put into a gift. Although the
common wisdom that “it’s the thought that counts” typically refers
to a receiver’s evaluations, existing psychological evidence led us
to predict that thoughts would count for gift givers by influencing
how connected they feel to the receiver. Thinking carefully about
another person can serve as a self-perception cue that makes a
person infer that he or she is more connected to someone else
(Bem, 1972). It also involves considering another’s perspective
carefully. Perspective taking tends to increase how closely con-
nected people feel to others (Davis et al., 1996; Galinsky et al.,
2005). We therefore predicted that gift givers would feel closer to
receivers after putting careful thought into a gift than after choos-
ing randomly. But because gift receivers do not experience the
effort of expending thoughts or the process of perspective taking,
we predicted that a gift giver’s thoughts would have no effect on
a gift receiver’s feelings of social connection. Because we ex-
pected that people would already feel very close to their friends
and family members, making it difficult to detect any experimen-
tally induced increase in social connection, we only investigated
gift exchanges between strangers in Experiment 4. Our theory
makes no prediction about how the quality of the gift would
influence a gift giver’s sense of connection, so we did not manip-
ulate whether the gift seemed relatively good or bad as in Exper-
iment 3.
Method
One hundred-fifty MSI visitors (93 female, one did not indicate
gender, M
age
⫽35.9 years) participated in a procedure identical to
Experiment 3 except that no comparison gift was used and that all
participants were paired with a stranger. After a brief introductory
meeting, participants were randomly assigned to be givers or
receivers and then separated into different rooms. Givers were
instructed to choose their gift either carefully or randomly, and
receivers were told in detail how givers were asked to choose.
After receiving their gift, receivers reported how much they
appreciated the gift, how grateful they felt, and how much they
liked the gift giver. Gift givers, in contrast, reported how much
appreciation, gratitude, and liking they expected the receiver of
their gift would feel. Both participants also reported their feelings
of social connection to their partner: how connected they felt, how
similar they felt, and how much they felt they understood the
other’s likes and dislikes. Gift givers also predicted how connected
their partner felt to them on the same measures. All responses were
made on scales ranging from 0 (not at all)to10(very much). Item
order (appreciation vs. connection) was counterbalanced but did
not influence any of the following analyses. As a manipulation
check, givers reported how much thought they expended and how
much they cared whether the receiver would like the gift on the
same 0 to 10 scale. Gift receivers answered the same two questions
in terms of how much thought and care they believed givers
expended.
Results and Discussion
Manipulation check. The thought manipulation appeared to
be effective. We averaged the two thought measures into a single
composite because they were highly correlated (r⫽.62). A 2
(role: giver vs. receiver) ⫻2 (thought: yes vs. no) mixed-model
ANOVA on the thought composite showed that the estimated
amount of thought was higher in the thoughtful condition than in
the thoughtless condition, F(1, 73) ⫽42.07, p⬍.01, ␩
p
2
⫽.37.
Specifically, givers reported that they expended more thoughts
when asked to choose thoughtfully (M⫽7.89, SD ⫽1.10) than
when asked to choose thoughtlessly (M⫽4.76, SD ⫽2.20), F(1,
73) ⫽59.70, p⬍.01, ␩
p
2
⫽.45. Receivers were well aware of this
difference, reporting that gift givers expended significantly more
thought in the thoughtful condition (M⫽6.22, SD ⫽1.97) than in
the thoughtless condition (M⫽5.07, SD ⫽2.49), F(1, 73) ⫽4.90,
p⫽.03, ␩
p
2
⫽.06. This main effect was qualified by a significant
676 ZHANG AND EPLEY
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

interaction between role and amount of thought, F(1, 73) ⫽8.97,
p⬍.01, ␩
p
2
⫽.11, indicating that the difference between the
thoughtful and thoughtless conditions was larger for givers than
for receivers.
Positive evaluation. We averaged the three positive evalua-
tion measures (appreciation, gratitude, and liking) into a single
composite (␣⫽.86). A 2 (role: giver vs. receiver) ⫻2 (thought:
yes vs. no) mixed-model ANOVA returned a significant main
effect of role, F(1, 73) ⫽38.48, p⬍.01, ␩
p
2
⫽.35, qualified by a
predicted interaction between role and thought, F(1, 73) ⫽6.27,
p⫽.01, ␩
p
2
⫽.08. As shown in Figure 3, the amount of thought
gift givers expended did not influence receivers’ positive evalua-
tion in any way (M
thoughtful
⫽7.22, SD ⫽1.86 vs. M
thoughtless
⫽
7.39, SD ⫽1.59), F⬍1. Gift givers, however, expected a more
positive evaluation when they chose thoughtfully (M⫽6.23,
SD ⫽1.79) than when they chose thoughtlessly (M⫽5.06, SD ⫽
1.90), F(1, 73) ⫽7.44, p⬍.01, ␩
p
2
⫽.09. Givers in this experi-
ment, like givers in previous experiments, did not predict correctly
when their thoughts would count.
An analysis on the main effect of role again demonstrated that
givers underestimated receivers’ positive evaluations (M
giver
⫽
5.63, SD ⫽1.93 vs. M
receiver
⫽7.30, SD ⫽1.72), both for
thoughtful gifts (M
giver
⫽6.23, SD ⫽1.79 vs. M
receiver
⫽7.22,
SD ⫽1.86), F(1, 73) ⫽6.75, p⫽.01, ␩
p
2
⫽.08, and for
thoughtless gifts (M
giver
⫽5.06, SD ⫽1.90 vs. M
receiver
⫽7.39,
SD ⫽1.59), F(1, 73) ⫽38.42, p⬍.01, ␩
p
2
⫽.34.
Social connection. We averaged the three social connection
measures (connection, similarity, and preference) into a single
composite (␣⫽.74). A 2 (role: giver vs. receiver) ⫻2 (thought:
yes vs. no) mixed-model ANOVA on the connection composite
measure returned a significant interaction between role and
thought, F(1, 73) ⫽4.24, p⬍.05, ␩
p
2
⫽.05. As predicted, givers
felt closer to receivers when they chose thoughtfully (M⫽3.70,
SD ⫽1.79) than when they chose thoughtlessly (M⫽2.62, SD ⫽
1.44), F(1, 73) ⫽8.29, p⬍.01, ␩
p
2
⫽.10. However, the givers’
thoughts had no influence on the receivers’ feelings of connection
(M
thoughtful
⫽3.31, SD ⫽1.81 vs. M
thoughtless
⫽3.41, SD ⫽2.08),
F⬍1. Thoughts did indeed count in this experiment, but for the
person having the thoughts rather than the person receiving the
gift.
Gift givers would appear to be surprised that receivers did not
feel closer to them when they chose carefully. A 2 (role: giver vs.
receiver) ⫻2 (thought: yes vs. no) mixed-model ANOVA com-
paring a receiver’s connection predicted by givers and a receiver’s
actual connection returned a significant interaction between role
and thought, F(1, 73) ⫽3.59, p⫽.06, ␩
p
2
⫽.05. Givers predicted
that the receivers’ evaluations would match their own, expecting
that receivers would feel more socially connected when they
chose thoughtfully (M⫽3.38, SD ⫽1.94) than when they chose
thoughtlessly (M⫽2.34, SD ⫽1.61), F(1, 73) ⫽6.24, p⫽.01,
␩
p
2
⫽.08. Receivers, however, did not feel significantly more
connected to givers who chose thoughtfully (M⫽3.31, SD ⫽
1.81) than to givers who chose thoughtlessly (M⫽3.41, SD ⫽
2.08), F⬍1. Gift givers expected that their thoughtfulness would
also make receivers feel more socially connected. They did not.
General Discussion
Choosing the perfect gift is difficult, involving costs of time,
thought, and money. Gifts are meant to be liked but also appreci-
ated in a way that strengthens a relationship. Putting a lot of
thought into getting just the right gift would seem to count con-
siderably for receivers and increase appreciation, because such
thoughts signal affection from the giver. It seems, after all, that
“it’s the thought that counts.” Our experiments suggest that this
common wisdom is exaggerated, mispredicted, and misplaced.
Exaggerated: The (Non)automaticity of Mental
State Inferences
The problem with counting thoughts in a gift exchange—or in
any other social context—is that thoughts are inherently invisible.
Because people do not have direct access to others’ thoughts,
considering them requires attention, effort, and motivation (Lin,
Keysar, & Epley, 2010). In the absence of any trigger to activate
mental state inferences, others’ thoughts are likely to remain
unconsidered and therefore have no impact on social evaluations.
In our experiments, a giver’s thoughts counted in receivers’ eval-
uations only when a friend gave a bad gift, the only condition we
believed would trigger receivers to consider a gift giver’s thoughts.
We observed this pattern both when gifts varied naturally in the
good versus bad gift conditions (Experiment 1 and 2) and when
gifts were held constant and manipulated only by comparison
(Experiment 3). In this way, the adage of “it’s the thought that
Figure 3. Ratings of positive evaluations (top panel) and social connec-
tion (bottom panel) by givers and receivers when a gift was chosen
thoughtfully or thoughtlessly (Experiment 4). Asterisks indicate a statisti-
cally significant simple effect between the thoughtful and thoughtless
conditions. Vertical lines within and above the bars indicate standard
deviation.
677
HOW THOUGHTS COUNT?
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

counts” is exaggerated. Thoughts count only in particular circum-
stances and count for nothing in many others.
We believe these results provide an important theoretical con-
tribution to the theory of mind literature by demonstrating an
important moderator of its impact in social interactions. These
results suggest that mental state inference, or theory of mind
reasoning, is not automatic or even primary in social judgments,
but instead must be activated by the social context. This evidence
comes from a novel paradigm for this literature in which we
measured the extent to which inferences about another person’s
thoughts were related to other consequential social judgments, in
this case to feelings of appreciation and gratitude. Our procedure
follows the logic of a classic experiment (Strack, Martin, &
Schwarz, 1988) in which participants were asked,“How happy you
are with your life in general?” and “How many dates did you have
last month?” When the happiness question preceded the dating
question, the two items were uncorrelated, r⫽⫺.12. However,
when the dating question preceded the happiness question, the
correlation was positive and significant, r⫽.66. All participants in
this experiment were perfectly able to think about the number of
dates they had and also considered this number relevant to their
happiness, but this life event did not spring automatically to mind
when assessing happiness unless these participants were explicitly
led to think of it.
Our results show a similar result for mental state inferences.
Although the gift receivers in our experiments were perfectly
capable of considering the thought a gift giver put into his or her
gift and applying the information to their evaluations, they did not
do so unless they were triggered to do so either implicitly (Exper-
iments 1 and 3) or explicitly (Experiment 2). In Experiment 2, we
demonstrated this most clearly by adopting the Strack et al. (1988)
procedure. When gift receivers reported their appreciation and
gratitude before considering a gift giver’s thoughts, there was no
relationship between these measures when recalling a good gift.
But when gift receivers reported their appreciation and gratitude
after being asked to consider a gift giver’s thoughts, there was a
strong relationship between a gift giver’s thoughts and a receiver’s
positive evaluations. Although this procedure does not measure the
activation of mental state reasoning directly, it does allow for a
direct manipulation of mental state reasoning. These results sug-
gest that failures to consider another person’s mental states may
not stem from an inability to do so or from a belief that another
person’s mental states are irrelevant, but rather stem from the lack
of sufficient motivation or cues to activate mental state reasoning.
Interestingly, those with high-functioning autism, usually con-
sidered to have a deficiency in the ability to think about the minds
of others, appear to show a similar pattern (Chevallier, Kohls,
Troiani, Brodkin, & Schultz, 2012). In one recent set of experi-
ments (Begeer, Malle, Nieuwland, & Keysar, 2010), people with
high-functioning autism were considerably less likely than normal
controls to describe another person’s actions using mental state
terms spontaneously. However, in a communication game in
which people were required to consider another person’s visual
perspective in order to perform the task (Keysar, Lin, & Barr,
2003), those with high-functioning autism showed no deficits in
mental state reasoning compared with healthy controls. These
results suggest that individual differences in mental state reasoning
may not stem from differences in the ability to think about the
mind of another person once an individual is trying to do so, but
rather from differences in the activation of mental state inferences.
Similar results have also been reported for the differences in
mental state reasoning between men and women. When given no
strong motivation to read another person’s thoughts accurately,
men tend to perform significantly worse than women. But when
triggered to read another person’s mind accurately by increasing
the incentives for doing so, men and women tend to perform
similarly (see Ickes, Gesn, & Graham, 2000, for a meta-analysis).
We think the important question for psychologists to address is
not who is better or worse at using their capacity to reason about
other minds, but rather when this capacity is activated. A full
understanding of mental state inferences, we believe (see also
Epley & Eyal, 2011), would consider four critical components:
activation (when people are triggered to reason about the minds of
others), application (how people reason about the minds of others
once they attempt to do so), accuracy (how well people reason
about others’ actual mental states), and adaptive functioning (how
the accuracy of mental state inferences is related to important
consequences). We believe the existing theory of mind has largely
overlooked the first stage of activation, often by suggesting that
mental state inferences are automatic or spontaneous and that this
activation stage is therefore unimportant.
We believe that is a mistake because it could exaggerate the
influence of mental state inferences in social interactions. We also
believe it is a mistake because it overlooks factors that moderate
the impact of mental state inferences. Our experiments investi-
gated the impact of one moderating factor—the extent to which a
gift is unexpected—on mental state inferences in evaluations. Our
experiments also suggested that contexts that make a gift giver’s
thoughtfulness more readily accessible could increase the impact
of mental state inferences. Gifts that make a giver’s thoughtfulness
relatively more transparent, such as highly personalized gifts, or
gifts that do not include a material object that could “crowd out”
mental state inferences, such as experiences rather than a material
possession, could increase the impact of a giver’s thoughts on
receiver’s evaluations.
Mispredicted: Mistaken Predictions by Gift Givers
Although gift givers anticipated the influence of gift quality on
receivers’ appreciations correctly, givers from all four experiments
consistently failed to predict correctly when their thoughts would
count and when they would not. Observer participants in the
pretest of Experiment 3 even predicted that a gift giver’s thoughts
would count uniformly across all conditions, overestimating quite
dramatically how much a gift giver’s thoughts actually count in
receivers’ evaluations. Because gift givers and outside observers
do not experience the same psychological processes that activate
mental state inferences among gift receivers, it can be challenging
to know when spending time and careful thought on a gift will
matter and when it will not. In this way, “it’s the thought that
counts” is mispredicted.
These results add to the large existing literature demonstrating
the difficulties people can have reading the minds of others accu-
rately (Epley, 2008; Ickes, 2003; Nickerson, 1999). We believe it
is interesting that gift givers were inconsistent across the experi-
ments in the extent to which they believed their thoughts would
count for gift receivers. In Experiment 1, gift givers appeared to
believe that gift receivers would appreciate a thoughtful gift more
678 ZHANG AND EPLEY
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

than a thoughtless gift, significantly so for a liked gift (r⫽.47)
and directionally so for a disliked gift (r⫽.28). We observed a
similar result in Experiment 4, where gift givers believed that
receivers would appreciate a gift chosen thoughtfully more than
one chosen thoughtlessly. Although results for gift givers in the
other experiments tended to show directionally similar effects,
they did not do so in all cases and certainly did not do so reliably.
Gift givers do not consistently think that thoughtful gifts will be
appreciated more than thoughtless gifts, but they sometimes think
so. The more distant observer participants in both the pretest
reported at the end of the introduction and the pretest to Experi-
ment 3, however, consistently showed in all conditions the con-
ventional wisdom that thoughts will count in gift exchanges.
We think these inconsistencies may stem from differences in the
natural accessibility of a gift giver’s thoughts versus the objective
quality of the gift in different contexts. Observers in Experiment 3,
for instance, did not actually have to go through the lengthy
process of choosing a gift and therefore did not consider the actual
quality of the gifts with great care. Instead, they only predicted the
impact of a thoughtful versus thoughtless decision on gift receiv-
ers’ evaluations in relative isolation. Similarly, in Experiment 4,
gift givers were not shown a comparison gift that would highlight
the relative desirability of a gift and instead were asked only to
choose either thoughtfully or thoughtlessly. Such contexts that
provide an isolated focus on a gift giver’s thoughts may therefore
increase their perceived importance in the social exchange. The
conventional wisdom that “it’s the thought that counts” may not
come from gift givers’ actual experience in social exchanges, but
rather may come from a natural focus on a gift giver’s thought-
fulness when these social exchanges are considered abstractly.
This suggestion is purely speculative, as we did not examine when
people would expect others to consider their thoughts and when
they would not.
Misplaced: Thoughts Count for Gift Givers
Although gift givers did not consistently think that their
thoughts would count for gift receivers, we did find evidence that
a gift giver’s thoughts counted in a different way for gift givers
themselves. In particular, we found that thoughts counted for gift
givers by making them feel more closely connected to gift receiv-
ers. This occurred, we believe, because putting careful thought into
a gift is an effortful process that serves as a signal to the quality of
a relationship for gift givers and is simultaneously an act of
perspective taking that previous research has suggested increases
relational bonds.
We believe these results for gift givers may shed some insight
into an important consequence of giving gifts to others. In partic-
ular, people report being significantly happier spending small
amounts of money on other people than spending the same amount
on themselves (Dunn, Aknin, & Norton, 2008). Our data suggest a
likely mechanism: that spending time thinking about a gift for
another person increases the sense of connection to that person.
Social connection is a reliable determinant of happiness (Argyle,
1999) and possibly the only necessary determinant (Diener &
Seligman, 2002), meaning that the sense of connection that comes
from expending thought on another person may be one cause of the
resulting happiness. Spending money on others may not be nec-
essary for increasing a giver’s happiness; happiness can come from
being thoughtful toward others. In this way, the common wisdom
that “it’s the thought that counts” is misplaced.
Conclusion
Although we have spent this General Discussion addressing the
theoretical implications of our research, any research on gift giving
naturally raises applied questions about giving better gifts. If
anything, our findings suggest that gift givers should give priority
to choosing gifts that receivers actually like rather than gifts that
reveal thoughtfulness. To the extent that gift givers falsely believe
that their thought processes count in a gift receiver’s evaluations,
they are likely to pay more attention to the symbolic meaning of an
exchange than they should. Although being thoughtful could lead
people to choose better gifts, it could also lead people to choose
worse gifts. For instance, when a gift is obviously good but does
not reveal a gift giver’s thoughtfulness, focusing on being thought-
ful could lead givers to choose a different gift in order to signal
their thoughtfulness. By doing so, gift givers could choose a
systematically worse gift. In one experiment (Gino & Flynn,
2011), participants believed that gifts chosen from a receiver’s
wish list would be considered less thoughtful and meaningful than
gifts that were not requested explicitly. This turned out to be
precisely wrong, and gift receivers actually appreciated receiving
a requested gift more than an unrequested yet thoughtful gift. Even
when thoughtfulness leads to choosing better gifts, it does not
increase gift receivers’ appreciation and gratitude beyond the qual-
ity of the gift itself. Gift receivers in our experiments did not
consistently appreciate thoughtful gifts. In fact, they only appre-
ciated thoughtful gifts if they were otherwise bad gifts.
Our research suggests different consequences of putting
thoughts into gift exchanges for givers and receivers. The practical
implications from this, we think, are fairly clear. If you want to
give a gift that someone will appreciate, then you should focus on
getting a good gift and ignore whether it is a thoughtful gift or not.
But if you want to feel closer to the person you are giving gifts to,
then put as much thought into your gift as you possibly can and do
not be offended when your thoughtfulness is overlooked.
References
Alicke, M. D. (2000). Culpable control and the psychology of blame.
Psychological Bulletin, 126, 556–574. doi:10.1037/0033-2909.126
.4.556
Ames, D., Flynn, F. J., & Weber, E. (2004). It’s the thought that counts: On
perceiving how favor-givers decide to help. Personality and Social
Psychology Bulletin, 30, 461–474. doi:10.1177/0146167203261890
Apperly, I. A., & Butterfill, S. A. (2009). Do humans have two systems to
track beliefs and belief-like states? Psychological Review, 116, 953–970.
doi:10.1037/a0016923
Argyle, M. (1999). Causes and correlates of happiness. In D. Kahneman, E.
Diener, & N. Schwarz (Eds.), Well-being: The foundations of hedonic
psychology, (pp. 353–373). New York, NY: Russell Sage Foundation.
Back, E., & Apperly, I. A. (2010). Two sources of evidence on the
non-automaticity of true and false belief ascription. Cognition, 115,
54–70. doi:10.1016/j.cognition.2009.11.008
Bargh, J. A., Chaiken, S., Govender, R., & Pratto, F. (1992). The generality
of the automatic attitude activation effect. Journal of Personality and
Social Psychology, 62, 893–912. doi:10.1037/0022-3514.62.6.893
Begeer, S., Malle, B. F., Nieuwland, M. S., & Keysar, B. (2010). Using
theory of mind to represent and take part in social interactions: Com-
679
HOW THOUGHTS COUNT?
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

paring individuals with high-functioning autism and typically develop-
ing controls. European Journal of Developmental Psychology, 7, 104–
122. doi:10.1080/17405620903024263
Bem, D. J. (1972). Self-perception theory. In L. Berkowitz (Ed.), Advances
in experimental social psychology (Vol. 6, pp. 1–62). New York, NY:
Academic Press.
Birch, S. A. J., & Bloom, P. (2007). The curse of knowledge in reasoning
about false beliefs. Psychological Science, 18, 382–386. doi:10.1111/
j.1467-9280.2007.01909.x
Cheal, D. (1986). The social dimensions of gift behavior. Journal of
Social and Personal Relationships, 3, 423–439. doi:10.1177/
0265407586034002
Cheal, D. (1988). The gift economy. New York, NY: Routledge.
Chen, S., & Chaiken, S. (1999). The heuristic-systematic model in its
broader context. In S. Chaiken & Y. Trope (Eds.), Dual-process theories
in social psychology, (pp. 73–96). New York, NY: Guilford Press.
Chevallier, C., Kohls, G., Troiani, V., Brodkin, E. S., & Schultz, R. T.
(2012). The social motivation theory of autism. Trends in Cognitive
Sciences, 16, 231–239. doi:10.1016/j.tics.2012.02.007
Cohen, A. S., & German, T. C. (2009). Encoding of others’ beliefs without
overt instruction. Cognition, 111, 356–363. doi:10.1016/j.cognition
.2009.03.004
Critcher, C. R., & Gilovich, T. (2010). Inferring attitudes from mindwan-
dering. Personality and Social Psychology Bulletin, 36, 1255–1266.
doi:10.1177/0146167210375434
Davis, M. H., Conklin, L., Smith, A., & Luce, C. (1996). Effect of
perspective taking on the cognitive representation of persons: A merging
of self and other. Journal of Personality and Social Psychology, 70,
713–726. doi:10.1037/0022-3514.70.4.713
Diener, E., & Seligman, M. E. (2002). Very happy people. Psychological
Science, 13, 81–84. doi:10.1111/1467-9280.00415
Duckworth, K. L., Bargh, J. A., Garcia, M., & Chaiken, S. (2002). The
automatic evaluation of novel stimuli. Psychological Science, 13, 513–
519. doi:10.1111/1467-9280.00490
Dunn, E. W., Aknin, L. B., & Norton, M. I. (2008, March 21). Spending
money on others promotes happiness. Science, 319, 1687–1688. doi:
10.1126/science.1150952
Epley, N. (2008). Solving the (real) other minds problem. Social & Per-
sonality Psychology Compass, 2, 1455–1474. doi:10.1111/j.1751-
9004.2008.00115.x
Epley, N., Caruso, E. M., & Bazerman, M. H. (2006). When perspective
taking increases taking: Reactive egoism in social interaction. Journal of
Personality and Social Psychology, 91, 872–889. doi:10.1037/0022-
3514.91.5.872
Epley, N., & Eyal, T. (2011). Integrations need both breadth and depth:
Commentary on Zaki and Ochsner. Psychological Inquiry, 22, 187–192.
doi:10.1080/1047840X.2011.567961
Epley, N., Keysar, B., Van Boven, L., & Gilovich, T. (2004). Perspective
taking as egocentric anchoring and adjustment. Journal of Personality
and Social Psychology, 87, 327–339. doi:10.1037/0022-3514.87.3.327
Epley, N., Morewedge, C., & Keysar, B. (2004). Perspective taking in
children and adults: Equivalent egocentrism but differential correction.
Journal of Experimental Social Psychology, 40, 760–768. doi:10.1016/
j.jesp.2004.02.002
Epley, N., & Waytz, A. (2010). Mind perception. In S. T. Fiske, D. T.
Gilbert, & G. Lindsay (Eds.), The handbook of social psychology (5th
ed., pp. 498–541). New York, NY: Wiley.
Fazio, R. H. (2001). On the automatic activation of associated evaluations:
An overview. Cognition and Emotion, 15, 115–141.
Flynn, F. J., & Brockner, J. (2003). It’s different to give than to receive:
Asymmetric reactions of givers and receivers to favor exchange. Journal
of Applied Psychology, 88, 1034–1045. doi:10.1037/0021-
9010.88.6.1034
Frith, C. D., & Frith, U. (2012). Mechanisms of social cognition. Annual
Review of Psychology, 63, 287–313. doi:10.1146/annurev-psych-
120710-100449
Galinsky, A. D., Ku, G., & Wang, C. (2005). Perspective-taking: Increas-
ing social bonds and facilitating social coordination. Group Processes &
Intergroup Relations, 8, 109–124. doi:10.1177/1368430205051060
Galinsky, A. D., Wang, C., & Ku, G. (2008). Perspective-takers behave
more stereotypically. Journal of Personality and Social Psychology, 95,
404– 419. doi:10.1037/0022-3514.95.2.404
Gilbert, D. T., & Hixon, J. G. (1991). The trouble of thinking: Activation
and application of stereotypic beliefs. Journal of Personality and Social
Psychology, 60, 509–517. doi:10.1037/0022-3514.60.4.509
Gino, F., & Flynn, F. (2011). Give them what they want: The benefits of
explicitness in gift exchange. Journal of Experimental Social Psychol-
ogy, 47, 915–922. doi:10.1016/j.jesp.2011.03.015
Gray, K., & Wegner, D. M. (2008). The sting of intentional pain. Psycho-
logical Science, 19, 1260–1262. doi:10.1111/j.1467-9280.2008.02208.x
Greenwald, A. G., Klinger, M. R., & Liu, T. J. (1989). Unconscious
processing of dichoptically masked words. Memory & Cognition, 17,
35–47. doi:10.3758/BF03199555
Guthrie, S. E. (1993). Faces in the clouds: A new theory of religion. New
York, NY: Oxford University Press.
Heider, F., & Simmel, M. (1944). An experimental study of apparent
behavior. The American Journal of Psychology, 57, 243–259. doi:
10.2307/1416950
Higgins, E. T. (1996). Knowledge activation: Accessibility, applicability,
and salience. In E. T. Higgins & A. W. Kruglanski (Eds.), Social
psychology: Handbook of basic principles (pp. 133–168). New York,
NY: Guilford Press.
Hogue, T. E., & Pebbles, J. (1997). The influence of remorse, intent and
attitudes toward sex offenders on judgments of a rapist. Psychology,
Crime, & Law, 3, 249–259. doi:10.1080/10683169708410821
Ickes, W. (2003). Everyday mind reading: Understanding what other
people think and feel. Amherst, NY: Prometheus Books.
Ickes, W., Gesn, P. R., & Graham, T. (2000). Gender differences in
empathic accuracy: Differential ability or differential motivation? Per-
sonal Relationships, 7, 95–109. doi:10.1111/j.1475-6811.2000
.tb00006.x
Keysar, B., & Barr, D. J. (2002). Self-anchoring in conversation: Why
language users do not do what they “should.” In T. Gilovich, D. Griffin,
& D. Kahneman, (Eds.), Heuristics and biases: The psychology of
intuitive judgment (pp. 150–166). New York, NY: Cambridge Univer-
sity Press.
Keysar, B., Lin, S., & Barr, D. J. (2003). Limits on theory of mind use in
adults. Cognition, 89, 25–41. doi:10.1016/S0010-0277(03)00064-7
Kleinke, C. L., Wallis, R., & Stadler, K. (1992). Evaluation of a rapist as
a function of expressed intent and remorse. The Journal of Social
Psychology, 132, 525–537. doi:10.1080/00224545.1992.9924732
Kunda, Z. (1999). Parallel processing of stereotypes and behaviors. In S.
Chaiken & Y. Trope (Eds.), Dual-process theories in social psychology
(pp. 314–322). New York, NY: Guilford Press.
Lin, S., Keysar, B., & Epley, N. (2010). Reflexively mindblind: Using
theory of mind to interpret behavior requires effortful attention. Journal
of Experimental Social Psychology, 46, 551–556. doi:10.1016/
j.jesp.2009.12.019
Malle, B. F., & Holbrook, J. (2012). Is there a hierarchy of social
inferences? The likelihood and speed of inferring intentionality,
mind, and personality. Journal of Personality and Social Psychology,
102, 661–684.
Morewedge, C. K. (2009). Negativity bias in attribution of external agency.
Journal of Experimental Psychology: General, 138, 535–545. doi:
10.1037/a0016796
Mussweiler, T. (2003). Comparison processes in social judgment: Mech-
anisms and consequences. Psychological Review, 110, 472–489. doi:
10.1037/0033-295X.110.3.472
680 ZHANG AND EPLEY
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Nickerson, R. S. (1999). How we know—and sometimes misjudge—what
others know: Imputing one’s own knowledge to others. Psychological
Bulletin, 125, 737–759. doi:10.1037/0033-2909.125.6.737
Pittman, T. S., & Pittman, N. L. (1980). Deprivation of control and the
attribution process. Journal of Personality and Social Psychology, 39,
377–389. doi:10.1037/0022-3514.39.3.377
Scholl, B. J., & Leslie, A. M. (1999). Modularity, development and ‘theory
of mind’. Mind & Language, 14, 131–153. doi:10.1111/1468-
0017.00106
Sperber, D., & Wilson, D. (2002). Pragmatics, modularity and mindread-
ing. Mind & Language, 17, 3–23. doi:10.1111/1468-0017.00186
Stone, V. E., Baron-Cohen, S., & Knight, R. T. (1998). Frontal lobe
contributions to theory of mind. Journal of Cognitive Neuroscience, 10,
640– 656. doi:10.1162/089892998562942
Strack, F., Martin, L. L., & Schwarz, N. (1988). Priming and communi-
cation: Social determinants of information use in judgments of life
satisfaction. European Journal of Social Psychology, 18, 429– 442.
doi:10.1002/ejsp.2420180505
Swim, J. K., Scott, E., Sechrist, G. B., Campbell, B., & Stangor, C. (2003).
The role of intent and harm in judgments of prejudice. Journal of
Personality and Social Psychology, 84, 944–959. doi:10.1037/0022-
3514.84.5.944
Uleman, J. S., Saribay, S. A., & Gonzalez, C. M. (2008). Spontaneous
inferences, implicit impressions, and implicit theories. Annual Review of
Psychology, 59, 329–360. doi:10.1146/annurev.psych.59.103006
.093707
Van Overwalle, F., Van Duynslaeger, M., Coomans, D., & Timmermans,
B. (2012). Spontaneous goal inferences are often inferred faster than
spontaneous trait inferences. Journal of Experimental Social Psychol-
ogy, 48, 13–18. doi:10.1016/j.jesp.2011.06.016
Waytz, A., Morewedge, C. K., Epley, N., Monteleone, G., Gao, J.-H., &
Cacioppo, J. T. (2010). Making sense by making sentient: Effectance
motivation increases anthropomorphism. Journal of Personality and
Social Psychology, 99, 410– 435. doi:10.1037/a0020240
Wimmer, H., & Perner, J. (1983). Beliefs about beliefs: Representation and
constraining function of wrong beliefs in young children’s understand-
ing of deception. Cognition, 13, 103–128. doi:10.1016/0010-
0277(83)90004-5
Wong, P. T., & Weiner, B. (1981). When people ask “why” questions, and
the heuristics of attributional search. Journal of Personality and Social
Psychology, 40, 650– 663. doi:10.1037/0022-3514.40.4.650
Young, L., & Saxe, R. (2009). An fMRI investigation of spontaneous
mental state inference for moral judgment. Journal of Cognitive Neuro-
science, 21, 1396–1405. doi:10.1162/jocn.2009.21137
Zaki, J., & Ochsner, K. (2011). Re-integrating the study of accuracy into
social cognition research. Psychological Inquiry, 22, 159–182. doi:
10.1080/1047840X.2011.551743
Received December 14, 2011
Revision received June 6, 2012
Accepted June 7, 2012 䡲
681
HOW THOUGHTS COUNT?
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.