When Does Feeling of Fluency Matter?

Abstract

It has been widely documented that fluency (ease of information processing) increases positive evaluation. We proposed and demonstrated in three studies that this was not the case when people construed objects abstractly rather than concretely. Specifically, we found that priming people to think abstractly mitigated the effect of fluency on subsequent evaluative judgments (Studies 1 and 2). However, when feelings such as fluency were understood to be signals of value, fluency increased liking in people primed to think abstractly (Study 3). These results suggest that abstract thinking helps distinguish central decision inputs from less important incidental inputs, whereas concrete thinking does not make such a distinction. Thus, abstract thinking can augment or attenuate fluency effects, depending on whether fluency is considered important or incidental information, respectively.

Full text

http://pss.sagepub.com/
Psychological Science
http://pss.sagepub.com/content/22/3/348
The online version of this article can be found at:
DOI: 10.1177/0956797611398494
2011 22: 348 originally published online 9 February 2011Psychological Science Claire I. Tsai and Manoj Thomas
When Does Feeling of Fluency Matter? : How Abstract and Concrete Thinking Influence Fluency Effects
Published by:
http://www.sagepublications.com
On behalf of:
Association for Psychological Science
can be found at:Psychological ScienceAdditional services and information for
http://pss.sagepub.com/cgi/alertsEmail Alerts:
http://pss.sagepub.com/subscriptionsSubscriptions:
http://www.sagepub.com/journalsReprints.navReprints:
http://www.sagepub.com/journalsPermissions.navPermissions:
at UNIV TORONTO MISSISSAUGA LIB on July 14, 2011pss.sagepub.comDownloaded from

Psychological Science
22(3) 348 –354
© The Author(s) 2011
Reprints and permission:
sagepub.com/journalsPermissions.nav
DOI: 10.1177/0956797611398494
http://pss.sagepub.com
People make judgments on the basis of not only information
but also the feelings they experience during judgment. In fact,
a substantial body of research has shown that feelings can
exert powerful influences on judgments (Pham, 2009;
Schwarz, 2004). We examined ways in which people respond
to an important class of feelings: fluency. We proposed that the
activation of abstract thinking can reduce the influence of feel-
ings on judgment, thereby attenuating and even reversing the
effects of fluency. Studies have shown that people consider
fluent (easy to process) stimuli more likeable, familiar, fre-
quent, true, and intelligent than similar but less fluent stimuli,
and those feelings of fluency affect judgment, even when the
judgment target has nothing to do with the actual causes of
fluency (Alter & Oppenheimer, 2009). One approach com-
monly adopted to mitigate such fluency effects is to prevent
people from misattributing fluency to value by drawing their
attention to the actual causes of fluency (Schwarz, 2004).
However, this debiasing approach can be difficult to general-
ize to real-world settings because it is not feasible to identify
the causes of fluency (or disfluency) whenever people experi-
ence it in everyday judgments.
We took a different approach to eliminating fluency effects,
namely, inducing abstract thinking (as opposed to concrete
thinking). Unlike judgment-specific moderators of fluency,
abstract and concrete thinking operate at the level of mind-sets
(Trope & Liberman, 2010) and can be induced using priming
tasks completely unrelated to judgment.1 For example, abstract
thinking can be enhanced by considering abstract categories
for everyday objects: Soda can be construed as a beverage,
chocolate can be construed as a sweet, and so forth. Con-
versely, concrete thinking can be enhanced by recalling con-
crete exemplars of objects: An example of soda is Coke, an
example of chocolate is dark chocolate, and so forth. The ease
with which these mind-sets can be activated suggests that
implementing them beyond laboratory studies should be rela-
tively easy.
Prior research has suggested that abstract thinking evokes
schematic processing and helps people focus on the big picture
by putting information in a larger perspective (Shanks &
Darby, 1998; Trope & Liberman, 2003, 2010). Consequently,
abstract thinking distinguishes central decision inputs from
incidental inputs during judgment. Because subjective feel-
ings are generally considered less important than objective
Corresponding Author :
Claire I. Tsai, Rotman School of Management, University of Toronto,
105 St. George St., Toronto, Ontario, Canada M5S 3E6
E-mail: claire.tsai@rotman.utoronto.ca
When Does Feeling of Fluency Matter? How
Abstract and Concrete Thinking Influence
Fluency Effects
Claire I. Tsai1 and Manoj Thomas2
1Rotman School of Management, University of Toronto, and 2S.C. Johnson Graduate School of Management, Cornell University
Abstract
It has been widely documented that fluency (ease of information processing) increases positive evaluation. We proposed and
demonstrated in three studies that this was not the case when people construed objects abstractly rather than concretely.
Specifically, we found that priming people to think abstractly mitigated the effect of fluency on subsequent evaluative judgments
(Studies 1 and 2). However, when feelings such as fluency were understood to be signals of value, fluency increased liking in
people primed to think abstractly (Study 3). These results suggest that abstract thinking helps distinguish central decision inputs
from less important incidental inputs, whereas concrete thinking does not make such a distinction. Thus, abstract thinking
can augment or attenuate fluency effects, depending on whether fluency is considered important or incidental information,
respectively.
Keywords
abstract thinking, construal level, fluency, feelings, evaluation, charitable giving
Received 4/7/10; Revision accepted 9/13/10
Research Article
at UNIV TORONTO MISSISSAUGA LIB on July 14, 2011pss.sagepub.comDownloaded from

Abstract Thinking and Fluency 349
information (Hsee & Tsai, 2008), we expected abstract think-
ing to reduce the relevance of fluency as a decision input and
thus weaken the effects of fluency on judgment. In contrast,
concrete thinking focuses people’s attention on specific
details: They see the trees rather than the forest. Consequently,
concrete thinking makes no distinction between central deci-
sion inputs (e.g., information content) and incidental decision
inputs (e.g., subjective feelings) and allows fluency to be taken
into consideration in judgment. Because fluency is likely to be
interpreted as value, we expected concrete thinkers to like flu-
ent stimuli more than less fluent stimuli.
In sum, we hypothesized that fluency influences evaluative
judgments under the condition of concrete thinking but not of
abstract thinking. We tested this hypothesis in two ways. In
Study 1, we manipulated fluency by presenting a text stimulus
in a clear or an unclear font. Because presenting information in
an unclear font renders processing more difficult without
changing the information being conveyed, we expected that
abstract thinking would attenuate the effect of fluency. In
Study 2, we further demonstrated the moderating effect of
abstract thinking by asking participants to generate a few or
many supporting arguments for an evaluation. Participants
who were asked to list a larger number of supporting argu-
ments not only experienced more difficulty, but also generated
more information. Because abstract thinking increases reli-
ance on information content, we expected abstract thinkers’
evaluation to become more favorable as information content
increased, despite the accompanying increase in processing
difficulty. Finally, in Study 3, we obtained evidence for the
proposed mechanism underlying the moderating effect of
abstract thinking.
Study 1
Study 1 tested the hypothesis that inducing abstract thinking
will moderate the effect of processing fluency on judgments of
liking in a subsequent unrelated task.
Method
Seventy-one students (23 males, 48 females; age range =
16–37 years) from the University of Toronto participated and
received $5 as compensation. Study 1 employed a 2 (mind-set:
abstract thinking vs. concrete thinking) × 2 (fluency: blurry
advertisement vs. clear advertisement) between-subjects
design. The study consisted of two ostensibly unrelated tasks.
The first was a word-generation task that induced abstract or
concrete thinking. The second task involved reading an adver-
tisement for chocolate. This task enabled us to examine the
moderating role of abstract thinking.
The word-generation task has been used successfully in
previous studies to manipulate subjects’ degree of abstract
thinking (Fujita, Trope, Liberman, & Levin-Sagi, 2006). Par-
ticipants were presented with 39 words (e.g., “soda,” “com-
puter”). In the abstract-thinking condition, participants
generated a superordinate category label for each word by
answering the question, “_________ is an example of what?”
Participants in the concrete-thinking condition generated a
subordinate exemplar for each word by answering the ques-
tion, “An example of __________ is what?”
Next, in an ostensibly unrelated study, participants read
an advertisement for LeVour chocolate, which we adapted
from extant research (Labroo & Kim, 2009). Depending on
the experimental condition, participants were shown either a
blurry (difficult to process) advertisement or a clear (easy to
process) advertisement (see the Supplemental Material
available online for reproductions of the two stimuli).
Everything else about the advertisement was held constant
across conditions.
After reading the advertisement, participants rated how
much they wanted to eat the chocolate, how desirable it was,
and how tempting it was; they responded to each question
separately on a scale from 1 (not at all) to 7 (very). As manipu-
lation checks, participants rated how difficult it was to read the
advertisement, how effortful reading it was, and how clear the
text of the advertisement was (reverse-coded for analysis);
they rated each measure of difficulty separately on a scale
from 1 (not at all) to 7 (very). Finally, participants rated their
mood on a scale from 1 (very bad/very sad) to 7 (very good/
very happy) and indicated how much attention they had paid to
the advertisement on a scale from 1 (a little) to 7 (a lot).
Results
Responses to the three items that measured the difficulty of
reading the advertisement were averaged to form a difficulty
index (α = .90). A two-way analysis of variance (ANOVA)
confirmed that the blurry advertisement was more difficult to
read than the clear one (blurry: M = 5.07; clear: M = 2.62),
F(1, 67) = 72.53, p < .001. Neither the main effect of mind-set
nor the interaction of fluency and mind-set had a significant
effect on the difficulty index (Fs < 1). Mood and attention did
not differ across conditions (Fs < 1).
The three ratings of the LeVour chocolates were averaged
to form a liking index (α = .93). A two-way ANOVA on the
liking scores revealed only a significant interaction of mind-
set and fluency, F(1, 67) = 8.99, p = .004, ηp
2 = .12. As
expected, participants in the concrete-thinking condition liked
the chocolate more when the advertisement was easy to pro-
cess (M = 5.83, SD = 1.51) than when it was difficult to pro-
cess (M = 4.42, SD = 1.39), F(1, 67) = 8.95, p = .004, ηp
2 = .12.
By contrast, processing fluency did not affect liking in the
abstract-thinking condition (blurry: M = 5.38, SD = 1.01;
clear: M = 4.77, SD = 1.56), F(1, 67) = 1.59, p = .21, ηp
2 = .02
(see Fig. 1).
Mediation analyses were conducted separately for the
abstract-thinking and concrete-thinking conditions (Baron &
Kenny, 1986). As predicted, the difficulty index mediated the
effect of fluency on liking in the concrete-thinking condition
(Sobel z = 2.15, p = .03), but not in the abstract-thinking
at UNIV TORONTO MISSISSAUGA LIB on July 14, 2011pss.sagepub.comDownloaded from

350 Tsai, Thomas
condition (Sobel z = 1.24, p = .21). For the concrete-thinking
condition, regression analyses confirmed a significant effect
of fluency on liking for LeVour chocolate, β = 0.71, SE = 0.24,
t(34) = 2.72, p = .006; a significant effect of fluency on the
difficulty index, β = −1.46, SE = 0.19, t(34) = −7.88, p < .001;
and a significant effect of the difficulty index on liking, β =
−0.31, SE = 0.14, t(34) = − 2.23, p = .03. When liking was
regressed simultaneously on the difficulty index and on flu-
ency, the effect of fluency was reduced in significance (p =
.09). For the abstract-thinking condition, regression analyses
revealed that neither fluency nor the difficulty index had a sig-
nificant effect on liking (both ps > .20). However, fluency had
a significant effect on the difficulty index, β = 0.98, SE = 0.22,
t(33) = 4.47, p < .001, a result confirming that regardless of the
degree of abstract thinking, the blurry advertisement was more
difficult to read than the clear one. These results support our
hypothesis that fluency increases liking when viewers are
thinking concretely, but has no effect on liking when viewers
are thinking abstractly.
Study 2
Study 2 further demonstrated the moderating effect of
abstract thinking. We expected abstract thinking to reverse
the fluency effect when increased processing difficulty was
accompanied by increased supporting information for judg-
ment. Study 2 expanded our findings by using a different
procedure to alter mind-set. Abstract thinking and concrete
thinking differ not only in whether broad categories or spe-
cific exemplars, respectively, are considered, but also in the
differential emphasis placed on end states of actions (the why
of activities; abstract thinking) or the means by which actions
are accomplished (the how of activities; concrete thinking;
Trope & Liberman, 2010). Study 2 investigated whether
thinking about the why or how aspects of an activity would
moderate the fluency effect. To increase generalizability, we
used nonhedonic stimuli and a behavioral measure of liking
judgment (charitable giving).
Method
Ninety-six students from the University of Toronto partici-
pated. Study 2 employed a 2 (mind-set: abstract thinking vs.
concrete thinking) × 2 (fluency: two reasons vs. eight reasons)
between-subjects design. Donation for charity was the depen-
dent variable. Participants first completed a task that induced
either abstract or concrete thinking. Depending on the experi-
mental condition, participants were asked to write a detailed
description of either why or how to study for an important
exam; we confirmed in pretesting that this procedure success-
fully altered mind-set. All participants were asked to describe
an important class they were taking and were given an entire
page to write down either reasons for doing well on the final
exam (the why-focused abstract-thinking condition) or a
detailed step-by-step study plan (the how-focused concrete-
thinking condition; see Tsai & McGill, in press, for more
details of the procedure).
Next, participants worked on a supposedly unrelated task
involving conserving the habitats of polar bears (see the Sup-
plemental Material for information given to participants).
Depending on the experimental condition, participants gen-
erated either two reasons (easy) or eight reasons (difficult)
for donating in support of the polar bears. They were then
asked how much money they would donate. To give their
decision monetary consequences, we stated that 5 partici-
pants would be randomly selected to receive $10 as a bonus
and asked participants how much they would donate if they
received the bonus. To ensure participants understood that
the decision was real, we told them that we would deduct the
donated amount from the bonus (and we did do so). Finally,
as manipulation checks, participants rated on 7-point scales
how difficult it was to generate reasons to donate (1 = not at
all, 7 = very), how much attention they paid to the informa-
tion about the conservation efforts (1 = little, 7 = a lot), and
how the information made them feel (1 = makes me feel bad,
7 = makes me feel good).
Results
A two-way ANOVA revealed a significant main effect of flu-
ency on the difficulty of thought generation (eight reasons:
M = 6.16;. two reasons: M = 4.50), F(1, 92) = 17.74, p < .001.
This result confirmed that generating eight reasons for donat-
ing money was more difficult than generating two reasons for
donating money. Neither the main effect of mind-set on the
difficulty of thought generation nor the interaction of mind-set
1
2
3
4
5
6
7
Preference (Liking Judgments)
Easy (Standard Ad) Difficult (Blurry Ad)
Abstract Thinking Concrete Thinking
Construal Mind-Set
Fig. 1. Results from Study 1: participants’ mean liking for chocolate shown in
an advertisement as a function of whether the ad was easy or difficult to read
and whether participants had been induced to think abstractly or concretely.
Liking judgments were made on a scale from 1 (not at all) to 7 (very). Error
bars represent standard errors of the mean.
at UNIV TORONTO MISSISSAUGA LIB on July 14, 2011pss.sagepub.comDownloaded from

Abstract Thinking and Fluency 351
and fluency was significant (Fs < 1). Attention paid to and
feelings toward the advertisement did not differ across condi-
tions (Fs < 1).
As expected, a two-way ANOVA on donation revealed a
significant interaction of mind-set (abstract vs. concrete) and
fluency, F(1, 92) = 9.11, p = .003, ηp
2 = .09 (see Fig. 2). Other
main effects were not significant, Fs < 1. As expected, in the
concrete-thinking condition, participants who had to generate
only two reasons donated more (M = $5.56, SD = 3.08) than
participants who had to generate eight reasons (M = $3.68,
SD = 3.42), F(1, 92) = 4.99, p = .02, ηp
2 = .05. Conversely, in
the abstract-thinking condition, participants who had to gener-
ate eight reasons donated more money (M = $5.27, SD = 2.88),
despite the increased difficulty of their task, than participants
who had to generate only two reasons (M = $3.26, SD = 2.94),
F(1, 92) = 4.24, p = .04, ηp
2 = .04.
As in Study 1, results of mediation analyses showed
that processing difficulty mediated the fluency effect in the
concrete-thinking condition (Sobel z = 2.24, p = .02), but not
in the abstract-thinking condition (p > .60). For the concrete-
thinking condition, regression analyses confirmed a signifi-
cant effect of fluency on donation, β = 0.94, SE = 0.44, t(53) =
2.14, p = .03; a significant effect of fluency on difficulty of
thought generation, β = −0.77, SE = 0.27, t(53) = −2.87, p =
.006; and a significant effect of difficulty of thought genera-
tion on donation, β = −0.69, SE = 0.19, t(53) = −3.55, p = .001.
When donation was regressed simultaneously on difficulty of
thought generation and on fluency, the effect of fluency was
no longer significant (p = .30).
In sum, the results of Study 2 show that fluency increases the
amount of money people are willing to donate when they are
thinking concretely, and that information content, rather than
fluency, affects how much money people are willing to donate
when they are thinking abstractly. The results from Studies 1
and 2 also challenge the assumption that global, abstract pro-
cessing induces low-effort heuristic processing (e.g., Smith
et al., 2006). If abstract thinking reduces processing effort, the
effect of fluency should have been more pronounced in the
abstract-thinking condition than in the concrete-thinking condi-
tion. However, consistent with other studies suggesting that
global processing does not alter processing effort (Bless et al.,
1996; Trope & Liberman, 2010; Vallacher & Wegner, 1987),
our study shows that fluency has no effect when people think
abstractly, and this suggests that the effect of abstract thinking
operates independently of processing effort.
In our next study, we directly tested the mechanism pos-
ited to underlie the moderating effect of abstract thinking. A
key premise of our hypothesis was that abstract thinking
omits fluency from judgment because feelings (such as those
about how fluent information processing is) are considered
less relevant decision inputs than information content. How-
ever, concrete thinking does not distinguish central informa-
tion from incidental information, and thus fluency, rather
than being dismissed as irrelevant, is unconsciously misat-
tributed to the judgment variable and interpreted as value
under conditions of concrete thinking.2 We tested this prem-
ise by prioritizing decision inputs: We instructed participants
to focus on either feelings or information content. If our
assumption is correct, then instructing abstract thinkers to
focus on feelings because feelings are important decision
inputs should augment, rather than attenuate, fluency effects.
But the same manipulation may elicit a different response
from concrete thinkers. Explicitly highlighting the impor-
tance of feelings should make concrete thinkers aware of
their unconscious misattribution. This awareness is likely to
thwart the misattribution process and eliminate fluency
effects (Berkowitz, Jaffee, Jo, & Troccoli, 2000).
Consequently, when product information signaled value,
we expected to replicate the findings of Study 1: That is,
we expected fluency would increase liking in the concrete-
thinking condition but would have no effect on liking in the
abstract-thinking condition. By contrast, when feelings sig-
naled value, we expected the opposite pattern: Fluency would
increase liking in the abstract-thinking condition because feel-
ings would be perceived as a worthwhile source of informa-
tion, and, therefore, fluency would be interpreted as value.
Conversely, fluency would have no effect in the concrete-
thinking condition because drawing attention to reliance on
feelings would disrupt the misattribution process.
Study 3
Moving beyond statistical mediation analyses, Study 3 directly
tested the proposed mechanism underlying the moderating
effect of abstract thinking. We prioritized decision inputs for
participants by manipulating the source of value: product
information or feelings.
0
2
4
6
8
Abstract Thinking Concrete Thinking
Construal Mind-Set
Easy (2 Reasons) Difficult (8 Reasons)
Donation ( )
Fig. 2. Results from Study 2: mean amount of money that participants were
willing to donate as a function of mind-set (abstract vs. concrete) and fluency
(providing two vs. eight reasons for making donations). Error bars represent
standard errors of the mean.
at UNIV TORONTO MISSISSAUGA LIB on July 14, 2011pss.sagepub.comDownloaded from

352 Tsai, Thomas
Method
One hundred fourteen students from the University of Toronto
participated. We used a 2 (source of value: product informa-
tion vs. feeling) × 2 (mind-set: abstract thinking vs. concrete
thinking) × 2 (fluency: blurry advertisement vs. clear adver-
tisement) between-subjects design. As in Study 1, participants
first completed a word-generation task that induced either
abstract or concrete thinking. Next, participants evaluated the
chocolate advertisement from Study 1. Before they read the
LeVour advertisement, participants in the feeling condition
were told to focus on their instinctive feelings while evaluat-
ing the chocolate. Participants assigned to the product-
information condition were asked to focus on the content of
the advertisement (see the Supplemental Material available
online for instructions provided to participants in both condi-
tions). Participants then viewed either a clear or a blurry print
advertisement. The dependent measures were the same as in
Study 1.
Results
As expected, the blurry advertisement was rated as more dif-
ficult to read than the clear advertisement (blurry: M = 5.69;
clear: M = 2.56), F(1, 106) = 185.95, p < .001. A 2 × 2 × 2
ANOVA on the averaged liking ratings for LeVour chocolate
(α = .95) revealed a significant three-way interaction,
F(1, 106) = 5.49, p = .02, ηp
2 = .05. Specifically, when partici-
pants were told to focus on the content of the advertisement,
the results replicated the findings of Study 1 (see Fig. 3a). Flu-
ency increased liking in the concrete-thinking condition
(blurry: M = 4.36, SD = 1.27; clear: M = 5.64, SD = 0.89),
F(1, 106) = 5.38, p = .02, ηp
2 = .04, but had no effect in the
abstract-thinking condition (blurry: M = 5.04, SD = 1.01;
clear: M = 5.02, SD = 0.94), F < 1. However, when partici-
pants were told to focus on their feelings, the effect was
reversed: Fluency increased liking in the abstract-thinking
condition (blurry: M = 4.12, SD = 1.16; clear: M = 5.27, SD =
1.01), F(1, 106) = 4.29, p = .04, ηp
2 = .05, but had no effect in
the concrete-thinking condition (blurry: M = 4.98, SD = 1.77;
clear: M = 4.80, SD = 1.47), F < 1 (see Fig. 3b).
General Discussion
Our studies show that the activation of a general mind-set,
abstract thinking, can eliminate fluency effects on judgment
and real behavior. We also tested the mechanism underlying
the moderating effect of abstract thinking and demonstrated
that abstract thinking can attenuate fluency effects if feelings
are considered less relevant signals of value or can augment
fluency effects if feelings are considered more relevant signals
of value. By contrast, concrete thinking makes no distinction
between central and incidental information and thus leads peo-
ple to misattribute fluency to value, except when the misattri-
bution process becomes evident.
Our findings dovetail with recent work by Häfner and Stapel
(2010), which suggests that rather than signaling value, fluency
can increase the usability of ambiguous information for judg-
ment. Using rhymed information as a fluent stimulus, Häfner
and Stapel found that fluency had no effect on how people use
unambiguous information, and they concluded that this was
because processing unambiguous information was easy regard-
less of whether the information rhymed or not. By contrast, flu-
ency (rhyming) increased the usability of ambiguous information
b
1
2
3
4
5
6
7
Concrete Thinking
Preference (Liking Judgments)
Construal Mind-Set
Easy (Standard Ad) Difficult (Blurry Ad)
Abstract Thinking
a
1
2
3
4
5
6
7
Preference (Liking Judgments)
Concrete Thinking
Construal Mind-Set
Easy (Standard Ad) Difficult (Blurry Ad)
Abstract Thinking
Fig. 3. Results from Study 3: participants’ mean liking for chocolate shown in
an advertisement as a function of whether the ad was easy or difficult to read
and whether participants had been induced to think abstractly or concretely.
Results are shown for (a) participants who were instructed to focus on the
content of the ad and (b) participants who were instructed to focus on their
feelings about the chocolate. Liking judgments were made on a scale from
1 (not at all) to 7 (very). Error bars represent standard errors of the mean.
at UNIV TORONTO MISSISSAUGA LIB on July 14, 2011pss.sagepub.comDownloaded from

Abstract Thinking and Fluency 353
and polarized evaluations, making positive attributes more posi-
tive and negative attributes more negative.
We examined fluency from a different angle: Rather than
seeing how fluency affects information processing, we consid-
ered how abstract or concrete thinking moderated fluency
effects when fluency served as a source of information
(Schwarz & Clore 2007). We found that concrete thinking
treated central and incidental information as equally usable,
leading people to attribute fluency to value. By contrast,
abstract thinking treated incidental information as less usable
than central information and discounted fluency when feelings
were considered less important inputs than information con-
tent. However, when feelings were considered important, the
situation was reversed. Moreover, our work advanced the field
by examining fluency effects in the context of mind-sets,
which can be induced by tasks completely unrelated to the pri-
mary judgment—a distinction of substantive practical and
theoretic importance.
Further, our research has important implications for studies
on feeling-based judgment. Other classes of feelings—specific
emotions, affect, mood, and proprioception—can also influence
judgment. Given that the mechanism underlying judgments
made on the basis of fluency is often the same as the mechanism
underlying judgments made on the basis of other classes of feel-
ings (i.e., the feeling-as-information framework; Pham, 2009;
Schwarz & Clore, 2007), abstract thinking may produce parallel
effects across different classes of feelings. It would be fruitful to
further investigate the interplay between abstract thinking and
feeling-based judgment.
Our work suggests that abstract thinking may reduce judg-
ment biases arising from overuse of contextual information.
When judgment seems complicated, people tend to employ
mental shortcuts or heuristics (e.g., availability heuristic:
Tversky & Kahneman, 1974; affect heuristic: Slovic,
Finucane, Peters, & MacGregor, 2002). Abstract thinking may
help mitigate judgment biases by reducing reliance on contex-
tual information in the same way that it reduced reliance on
fluency in our studies. This is important because numerous
studies have shown that judgment biases are hard to eliminate
(Camerer & Hogarth, 1999). Our work suggests that activating
abstract thinking may be a promising debiasing strategy.
Our results also raise an interesting question concerning the
typical level of abstractness of thinking in previous fluency
studies. Prior research suggests that people generally process
information abstractly (Vallacher & Wegner, 1987), but they
are likely to think concretely and exhibit fluency effects when
performing unfamiliar tasks in laboratory experiments (e.g.,
making risk judgments about cardiovascular disease). This
proposition is consistent with prior findings that fluency
effects occur for unfamiliar tasks but not for familiar ones
(Haddock, Rothman, Reber, & Schwarz, 1999) and that unfa-
miliar or difficult tasks induce lower-level processing (Navon,
1977; Vallacher & Wegner, 1987). Although our results sup-
port these findings, further investigation is required to deter-
mine people’s typical level of abstraction and explore its
implications for fluency effects.
Acknowledgments
The two authors contributed equally to this research. The authors
gratefully acknowledge helpful comments from Ann McGill,
Chen-Bo Zhong, Cheryl Wakslak, and members of the Marketing
Brownbag Series (especially Scott Hawkins, Dilip Soman, and Min
Zhao) at the Rotman School of Management, University of Toronto.
Declaration of Conflicting Interests
The authors declared that they had no conflicts of interest with
respect to their authorship or the publication of this article.
Funding
This research was supported in part by a grant from the Desautels
Centre for Integrative Thinking at the Rotman School of Management.
Supplemental Material
Additional supporting information may be found at http://pss.sagepub
.com/content/by/supplemental-data
Notes
1. Drawing on prior work on procedural mind-set (Xu & Wyer, 2007),
we considered abstract thinking and concrete thinking to be processing
schema that operate at the level of mind-sets. Therefore, when abstract
thinking and concrete thinking are activated in one context, they influ-
ence subsequent judgments in completely unrelated contexts.
2. Several studies have shown that unconscious misattribution is
responsible for the effect of fluency on preferences (e.g., Menon &
Raghubir, 2003; Schwarz et al., 1991).
References
Alter, A.L., & Oppenheimer, D.M. (2009). Uniting the tribes of flu-
ency to form a metacognitive nation. Personality and Social Psy-
chology Review, 13, 219–235.
Baron, R.M., & Kenny, D.A. (1986). The moderator-mediator vari-
able distinction in social psychological research: Conceptual,
strategic and statistical considerations. Journal of Personality
and Social Psychology, 51, 1173–1182.
Berkowitz, L., Jaffee, S., Jo, E., & Troccoli, B.T. (2000). On the
correction of feeling-induced judgmental biases. In J.P. Forgas
(Ed.), Feeling and thinking: The role of affect in social cognition
(pp. 131–152). New York, NY: Cambridge University Press.
Bless, H., Clore, G.L., Schwarz, N., Golisano, V., Rabe, C., & Wölk,
M. (1996). Mood and the use of scripts: Does being in a happy
mood really lead to mindlessness? Journal of Personality and
Social Psychology, 71, 665–679.
Camerer, C., & Hogarth, R. (1999). The effects of financial incen-
tives in experiments: A review and capital-labor-production
framework. Journal of Risk and Uncertainty, 19, 7–42.
Fujita, K., Trope, Y., Liberman, N., & Levin-Sagi, M. (2006). Con-
strual levels and self-control. Journal of Personality and Social
Psychology, 90, 351–367.
Haddock, G., Rothman, A.J., Reber, R., & Schwarz, N. (1999). Form-
ing judgments of attitude certainty, importance, and intensity:
The role of subjective experiences. Personality and Social Psy-
chology Bulletin, 25, 771–782.
at UNIV TORONTO MISSISSAUGA LIB on July 14, 2011pss.sagepub.comDownloaded from

354 Tsai, Thomas
Häfner, M., & Stapel, D.A. (2010). Information to go: Fluency
enhances the usability of primed information. Journal of Experi-
mental Social Psychology, 46, 73–84.
Hsee, C.K., & Tsai, C.I. (2008). Hedonomics in consumer behavior.
In C. Haugtvedt, P. Herr, & F. Kardes (Eds.), Handbook of con-
sumer psychology (pp. 639–658). Mahwah, NJ: Erlbaum.
Labroo, A.A., & Kim, S. (2009). The “instrumentality” heuristic:
Why metacognitive difficulty is desirable during goal pursuit.
Psychological Science, 20, 127–134.
Menon, G., & Raghubir, P. (2003). Ease-of-retrieval as an automatic
input in judgments: A mere-accessibility framework? Journal of
Consumer Research, 30, 230–243.
Navon, D. (1977). Forest before trees: The precedence of global
features in visual perception. Cognitive Psychology, 9,
353–383.
Pham, M.T. (2009). The lexicon and grammar of affect as informa-
tion in consumer decision making: The GAIM. In M. Wänke
(Ed.), Social psychology of consumer behavior (pp. 167–200).
London, England: Psychology Press.
Schwarz, N. (2004). Metacognitive experiences in consumer judg-
ment and decision making. Journal of Consumer Psychology, 14,
332–348.
Schwarz, N., Bless, H., Strack, F., Klumpp, G., Rittenauer-Schatka,
H., & Simons, A. (1991). Ease of retrieval as information:
Another look at the availability heuristic. Journal of Personality
and Social Psychology, 61, 195–202.
Schwarz, N., & Clore, G.L. (2007). Feelings and phenomenal experi-
ences. In E.T. Higgins & A. Kruglanski (Eds.), Social psychology:
Handbook of basic principles (2nd ed., pp. 385–407). New York,
NY: Guilford.
Shanks, D.R., & Darby, R.J. (1998). Feature- and rule-based gener-
alization in human associative learning. Journal of Experimental
Psychology: Animal Behavior Processes, 24, 405–415.
Slovic, P., Finucane, M., Peters, E., & MacGregor, D.G.
(2002). The affect heuristic. In T. Gilovich, D. Griffin, &
D. Kahneman (Eds.), Heuristics and biases: The psychology of
intuitive judgment (pp. 397–420). New York, NY: Cambridge
University Press.
Smith, E.R., Miller, D.A., Maitner, A.T., Crump, S.A., Garcia-Marques,
T., & Mackie, D.M. (2006). Familiarity can increase stereotyping.
Journal of Experimental Social Psychology, 42, 471–478.
Trope, Y., & Liberman, N. (2003). Temporal construal. Psychological
Review, 110, 403–421.
Trope, Y., & Liberman, N. (2010). Construal-level theory of psycho-
logical distance. Psychological Review, 117, 440–463.
Tsai, C.I., & McGill, A.L. (in press). No pain, no gain? How con-
strual level and fluency affect consumer confidence. Journal of
Consumer Research.
Tversky, A., & Kahneman, D. (1974). Judgment under uncertainty:
Heuristics and biases. Science, 185, 1124–1130.
Vallacher, R.R., & Wegner, D.M. (1987). What do people think
they’re doing? Action identification and human behavior. Psy-
chological Review, 94, 3–15.
Xu, A.J., & Wyer, R.S. (2007). The effect of mindsets on con-
sumer decision strategies. Journal of Consumer Research, 34,
556–566.
at UNIV TORONTO MISSISSAUGA LIB on July 14, 2011pss.sagepub.comDownloaded from