Journal of Experimental Social Psychology 42 (2006) 627 631
www.elsevier.com/locate/jesp
On the beneWts of thinking unconsciously: Unconscious

thought can increase post-choice satisfaction
Å„
Ap Dijksterhuis , Zeger van Olden
Social Psychology Program, University of Amsterdam, Roetersstaat 15, 1018 WB, Amsterdam, The Netherlands
Received 22 May 2005; revised 7 September 2005
Available online 13 December 2005
Abstract
This work compares conscious thought and unconscious thought in relation to quality of choice. Earlier work [Dijksterhuis, A. (2004).
Think diVerent: The merits of unconscious thought in preference development and decision making. Journal of Personality and Social
Psychology, 87, 586 598] has shown that people make better choices after engaging in unconscious thought (i.e., unconscious activity dur-
ing a period of distraction) rather than in conscious thought. However, the evidence was obtained for choices between hypothetical alter-
natives with quality of choice operationalized normatively. As quality of decision is essentially subjective, in the current experiment
participants chose between real objects with quality operationalized as post-choice satisfaction. In a paradigm based on work by Wilson
and colleagues [Wilson, T. D., Lisle, D., Schooler, J. W., Hodges, S. D., Klaaren, K. J., & LaFleur, S. J. (1993). Introspecting about reasons
can reduce post-choice satisfaction. Personality and Social Psychology Bulletin, 19, 331 339], participants were brieXy presented with Wve
art posters, and chose one either (a) immediately, (b) after thorough conscious thinking about each poster, or (c) after a period of distrac-
tion. Participants took their favorite poster home and were phoned 3 5 weeks later. As hypothesized, unconscious thinkers were more sat-
isWed with their choice than participants in the other two conditions.
© 2005 Elsevier Inc. All rights reserved.
Keywords: Unconscious thought; Consciousness; Choice; Decision making
Imagine being at an art auction in Paris. There is a like or dislike certain aspects. You may even use a balance
Monet on sale for a mere 100 million Euros and a Van sheet and engage in assigning the paintings pluses and
Gogh for a steep 125 million. Unfortunately, you ve been a minuses for diVerent attributes. Such a strategy, we tend to
college professor rather than a Fortune 500 CEO all your think, leads to a decision we will be most happy with.
life, so your meager savings (about 150 million) only allow The past Wfteen years have witnessed a shift in our think-
you to purchase one. And let us assume it is a diYcult deci- ing. Variou s researchers have shown that it is not always
sion because you like both Monet and Van Gogh. What to beneWcial to think consciously about decisions (Claxton,
do? How to make this choice? 1997; Dijksterhuis, 2004; Dijksterhuis, Smith, van Baaren, &
Psychologists and lay people alike have long maintained Wigboldus, 2005; Levine, Halberstadt, & Goldstone, 1996;
that thorough conscious deliberation leads to the best deci- Pelham & Neter, 1995; Schooler, Ohlsson, & Brooks, 1993;
sions. When you face an important choice, think about it Simonson & Nowlis, 2000; Wilson & Schooler, 1991; Wilson
carefully. In the case of the choice between Monet and van et al., 1993). Conscious thought has shortcomings that can
Gogh, scrutinize both paintings, think thoroughly about prevent sound decision making. First of all, conscious
what you like and what you don t like, and about why you thought can lead to suboptimal weighting of the importance
of aspects of diVerent choice alternatives. In addition,
because consciousness has low capacity, conscious thought

This research was supported by NWO-Vernieuwingsimpuls 016.025.030.
* often leads people to take into account only a limited subset
Corresponding author. Fax: [phone].
of information at the expense of other information that
E-mail address: [e-mail] (A. Dijksterhuis).
0022-1031/$ - see front matter © 2005 Elsevier Inc. All rights reserved.
doi:10.1016/j.jesp.2005.10.008
628 A. Dijksterhuis, Z. van Olden / Journal of Experimental Social Psychology 42 (2006) 627 631
should be taken into account when making a decision de Vries, 2005) shows that such unconscious thought is goal
(Dijksterhuis, 2004; Wilson & Schooler, 1991). directed. In their experiment, participants Wrst read infor-
Should this lead us to abandon conscious thought and to mation pertaining to a decision problem. Subsequently, all
start making decisions without any thought at all? Is it a participants were distracted, but some were told that they
good idea to just ask the people at the auction to wrap the would later have to choose between the various decision
van Gogh after having merely glanced at both paintings? alternatives, whereas others were told they could forget the
The answer is no. In addition to conscious thought, there is information and that they would not be asked any ques-
a second mode of thought, that we (Dijksterhuis, 2004; tions about it. As it turned out, the former group showed
Dijksterhuis & Nordgren, in press) called unconscious superior memory for the information, and indeed showed
thought. Unconscious thought refers to unconscious pro- better organization of information in memory. The latter
cesses dealing with a problem while consciousness is group did not demonstrate any meaningful memory orga-
directed elsewhere. It is thought without conscious atten- nization. In sum, only participants who knew they had to
tion and related to the notion of incubation in creativity choose between the apartments sometime later engaged in
research, and the idea to investigate unconscious thought unconscious thought.
was indeed partly based on research on creativity and These and other (e.g., Dijksterhuis & Meurs, in press) Wnd-
related domains. Bowers, Regehr, Balthazard, and Parker ings have led to the development of the Unconscious
(1990) have shown that when people solve a puzzle, the Thought Theory (UTT; Dijksterhuis & Nordgren, in press).
unconscious makes progress at Wnding a solution without This theory describes various characteristics of both con-
any conscious awareness of this progress. Yaniv and Meyer scious and unconscious thought on the basis of which spe-
(1987) found that when people are consciously trying to ciWc hypotheses can be derived that are applicable to decision
retrieve a word they feel they know (the  tip-of-the-tongue making, attitude formation, impression formation, and crea-
phenomenon), the unconscious activation of the word tivity. For example, whereas unconscious thought works
increases long before any conscious awareness. In these  bottom-up and can integrate large amounts of informa-
studies, we may say that people engaged in unconscious tion, conscious thought is very limited in its capacity and
thought. works  top-down. One of the central principles of the
Betsch, Plessner, Schwieren, and Gütig (2001) demon- model is the observation that consciousness has limited
strated that unconscious thought can even deal with rela- capacity. This limited capacity of consciousness had led to
tively complex judgments. In their experiments, the the hypothesis that conscious thought often leads to rela-
unconscious was shown to be able to integrate large tively poor decisions. Indeed, the experiments by Dijksterhuis
amounts of information without conscious guidance. Their (2004) earlier discussed conWrmed this hypothesis.
participants were presented with ads at a computer screen Although more research is needed to understand exactly
while simultaneously numerical increases and decreases of how unconscious thought renders superior decisions, some
Wve hypothetical shares were shown. Participants were can be speciWed. First, the unconscious has very high capac-
requested to focus on the ads (they knew they had to ity, leading unconscious thought to take into account all
answer questions about the ads later on), but afterwards information rather than just a subset. Second, unconscious
they had developed a  gut feeling towards the shares. thought works  bottom-up and weights the relative impor-
They somehow knew what the good and bad shares were, tance of diVerent attributes of objects in a relatively objective
without being able to verbalize why. These results are and  natural way. During unconscious thought, diVerent
intriguing. In the absence of conscious attention, people attributes are weighted and evaluated and then integrated
somehow Wgured out what the best share was. into an overall  evaluative summary judgment of all rele-
Recently, Dijksterhuis (2004) tested the hypothesis that vant information. Conversely, conscious thought often dis-
unconscious thought can lead to superior decisions relative turbs this natural weighting process (see also Wilson et al.,
to conscious thought. In various experiments, participants 1993) and can therefore lead to inferior decisions.
made decisions under three diVerent experimental condi- However, it is warranted to further investigate this
tions. After participants had read information about a deci- hypothesis as the experiments by Dijksterhuis (2004) have
sion problem (i.e., information about three diVerent two important limitations. In the experiments participants
apartments), they either had to decide immediately, after a were presented with hypothetical alternatives (e.g., apart-
few minutes of conscious thought about the choice alterna- ments), whereas quality of decision was operationalized nor-
tives, or after a few minutes of distraction. Unconscious matively. One of the alternatives was made more desirable
thinkers (i.e., distracted people) made superior decisions rel- than others by giving it more positive than negative aspects.
ative to participants in the other two conditions. It was also However, quality of decision is essentially subjective, and the
shown that the superior decisions made by unconscious Wndings by Dijksterhuis (2004) beg for a replication with a
thinkers was caused by a productive thought process, subjective operationalization of quality of choice. Hence, the
rather than by mere distraction. For instance, it was dem- goal of the current experiment is to investigate whether the
onstrated that unconscious thought renders information to beneWcial eVects of unconscious thought generalize across
become better organized in memory (as assessed by cluster- more realistic choices, whereby quality of decision is opera-
ing scores). Moreover, recent evidence (Dijksterhuis & tionalized as post-choice satisfaction.
A. Dijksterhuis, Z. van Olden / Journal of Experimental Social Psychology 42 (2006) 627 631 629
In testing our hypothesis, we will use a paradigm devel- were requested to indicate which one they liked most. In
oped by Wilson et al. (1993) who investigated the eVects of the Conscious thought condition, each poster appeared on
conscious thought on post-choice satisfaction. In their the screen individually for 90 s. Participants were asked
experiment, participants chose between art posters under to look at each poster again carefully, and to list reasons
two diVerent conditions. Some participants were asked to for why they liked or disliked each poster and to carefully
analyze their thoughts by listing their reasons for liking or analyze their preferences. Participants were given pen
disliking each poster. Control participants, on the other and paper to list their thoughts. After participants ana-
hand, were not given the opportunity to consciously think lyzed their evaluation of the various posters, all posters
before they chose. All participants were given their favorite appeared on the screen simultaneously and participants
art poster to take home and were phoned a few weeks later. were asked to indicate which one they liked most. In the
Participants who consciously analyzed their choices were Unconscious thought condition, participants were told
less satisWed with their posters than control participants. that they would engage in another task for a while, after
Whereas Wilson et al. (1993) tested and conWrmed the which they would be asked which poster they liked best.
hypothesis that conscious thought would lead to poor deci- These participants solved anagrams for 450 s (i.e., the
sions compared to no conscious thought, we test the same time the conscious thinkers analyzed), after which
hypothesis that unconscious thought leads to superior deci- all posters appeared on the screen simultaneously again
sions. Testing this hypothesis entails the use of a condition and participants were asked to indicate which one they
in which participants have no time to think at all, and a liked most.
condition in which participants are temporarily distracted After participants had indicated their preference, they
to engage in unconscious thought. As we include a con- were asked for their attitude towards each poster. We asked
scious thought condition as in the study by Wilson et al. for liking ratings and used scales with only the poles labeled
(1993), in the current experiment we run three conditions ( not at all and  very much ). Between the poles a line
(as in Dijksterhuis, 2004). Our baseline condition is an was drawn and participants could indicate their attitude by
immediate decision condition, in addition we include a con- clicking somewhere on this line. In reality, we used a 50-
scious thought condition and an unconscious thought con- points scale (from Ä„25 to 25). Subsequently, they were
dition. Our hypothesis entails that unconscious thinkers asked how knowledgeable they were about art, again on a
make better decisions (show more post-choice satisfaction) 50-points scale. Finally, participants were asked to give
than participants in the other two conditions. their phone numbers in case  something goes wrong with
data storage. Participants were then requested to return to
Method the experimenter who awaited them with a surprise: the
participants could take their favorite poster home. Subse-
Participants and design quently, participants were (partly) debriefed, thanked, and
dismissed.
One-hundred-and-thirteen undergraduate students (87 Three to Wve weeks later participants were phoned with
women and 26 men) of the University of Amsterdam were the question whether they could remember participating in
randomly assigned to one of three conditions: an immedi- the experiment where they had received a poster. The exper-
ate decision condition; a conscious thought condition; and imenter continued by asking participants how satisWed they
an unconscious thought condition. They either received were with the poster they chose (on a 10-point scale from 1
course credits or 7 Euros. tot 10). Second, the experimenter asked how much regret
they experienced after their choice (again, on a 10 point
Materials and procedure scale from 1 tot 10), and Wnally, the experimenter asked
them how many Euros they would be willing to sell their
Participants were seated in individual cubicles in front poster for. Participants were then thanked and, for those
of a computer. The computer program provided the who wanted, fully debriefed.
instructions. The experiment was introduced as being on
 Visual preferences and the evaluation of art. After Results
administering a few standard demographic questions, par-
ticipants were asked to look at Wve diVerent digital images Of the 113 participants, 24 could not be reached over the
of art posters. Three out of the Wve posters were abstract phone, leaving 89 participants available for the analyses. It
art posters, whereas the two others depicted photographs was Wrst established that there were no signiWcant diVer-
(of a Xower and of a Xock of birds). The images were pre- ences of choice of posters between conditions.
sented randomly and each image appeared on the com-
puter screen for 15 s. After presentation of the Main dependent variables
photographs, the computer assigned participants to one
of three conditions. Satisfaction
In the Immediate decision condition, all Wve posters The satisfaction scores were subjected to an analysis of
appeared on the screen simultaneously and participants covariance with Condition and as a factor and Knowledge
630 A. Dijksterhuis, Z. van Olden / Journal of Experimental Social Psychology 42 (2006) 627 631
Table 1
positively than others. However, these strong preferences
Satisfaction, amount of money asked for selling the poster (in Euros), and
are sometimes simply wrong, because people use the wrong
strength of preference for all conditions (standard deviations between
 weighting schemes. They attach too much or too little
parentheses)
weight to various aspects of the stimuli they have to choose
Condition Immediate Conscious Unconscious
from (see also Dijksterhuis, 2004; Schooler et al., 1993). Our
Satisfaction 6.68 (1.28) 6.68 (.88) 7.30 (.88)
data may conWrm the ideas of Wilson and colleagues.
Money 6.39 (5.94) 5.03 (5.20) 9.56 (8.68)
Strength of preference 17.8 (6.0) 20.7 (6.4) 15.8 (6.0)
Strength of preference
of art as a covariate1. The eVect of Condition was signiW-
The sum of the attitudes towards the four posters not cho-
cant, F(2, 85)D3.71, p <.03, 2D.08. As can be seen in Table
sen were subtracted from the attitude towards the chosen
1, unconscious thinkers were more satisWed with their
poster. The resulting diVerence scores represent how strong
choice than conscious thinkers, F(1, 58)D8.96, p < .005,
participants preferences are. The scores were subjected to an
2D.13, and than immediate decision makers,
analysis of covariance with Condition as a factor and
F(1, 52)D4.14, p < .05, 2D.07. These latter two conditions
Knowledge of art as a covariate. The eVect of Condition was
did not diVer.
signiWcant, F(2,85)D6.58, p < .002, 2D.13, and supported
the theorizing by Wilson and colleagues. As can be seen in
Regret
Table 1, conscious thinkers had stronger, rather than weaker
The same analysis of covariance yielded no eVects on
preferences than unconscious thinkers, F(1,58)D11.55,
regret (all F s < 1.1). Participants indicated experiencing
p <.001, 2D.17. The strength of preference among the
hardly any regret (all cell means below 2).
immediate decision makers fell in between the two other con-
ditions and only diVered signiWcantly from the unconscious
Money
thought condition, F(1,52)D5.63, p< .03, 2D.10.
The data on the amount of money participants were pre-
pared to sell their poster for were also subjected to the same
Correlations between main DV s and attitudes towards
analysis of covariance. The eVect of Condition was signiW-
chosen poster
cant, F(2, 85)D3.24, p <.05, 2D.07. As can be seen in Table
1, participants in the unconscious thought condition
Given that conscious thinkers have strong preferences
wanted more money for their poster than participants in
but report less satisfaction (at least compared to uncon-
the conscious thought condition, F(1, 58)D6.65, p <.02,
scious thinkers), may we draw the conclusion that their
2D.12, and than participants in the immediate decision
preferences are wrong? To shed light on this is we corre-
condition, although this eVect failed to reach signiWcance,
lated the attitudes towards the chosen poster with the
F(1, 52)D2.28, p < .14, 2D.04. These latter two conditions
main DV s. SpeciWcally, the attitude scores were, for each
did not diVer.
condition separately, correlated with satisfaction, regret,
and amount of money participants were willing to sell
Attitudes
their poster for. These correlations did not diVer signiW-
cantly between conditions, but a trend became apparent.
After having chosen a poster in the laboratory session,
Whereas none of the correlations for the conscious think-
participants also indicated their attitudes towards each of
ers reached signiWcance, two out of the three were signiW-
the individual posters. With these data we may obtain a
cant (and in the predicted direction) in the other two
better understanding of the nature of the decision process
conditions. For immediate decision makers, attitudes
in the various experimental conditions. It is possible that
towards the chosen poster predicted satisfaction (rD.61,
conscious reasoning disturbs the decision process because it
p < .001) and  money (rD.48, p < .01). For unconscious
leads people have less strong preferences. However, it is
thinkers, attitudes towards the chosen poster predicted
also possible that conscious thinkers did have strong pref-
satisfaction (rD.40, p < .05) and regret (rDÄ„.46, p <.03).
erences, but that their preferences were suboptimal. Wilson,
In sum, attitudes were predictive of post-choice satisfac-
Schooler, and colleagues (Wilson et al., 1993; Wilson &
tion for immediate decision makers and for unconscious
Schooler, 1991) have argued in favor of the second possibil-
thinkers, whereas such evidence was not obtained form
ity. Conscious thinkers do have strong preferences, in the
conscious thinkers. Again however, the correlations did
sense that one attitude object is clearly evaluated more
not diVer between conditions and should not be overinter-
preted.
1
The choice to include knowledge of art as a covariate was based on the
Wilson et al. (1993) experiment in which the same was done. The covariate
General discussion
did not regress signiWcantly on any of the main dependent variables or the
attitudes towards the chosen poster (all p s > .05) although the F values
People who were given the opportunity to think about
were often greater than 1. However, knowledge of art did have a signiW-
choices unconsciously made superior decisions relative to
cant eVect on preferences (p < .01). As one may expect, more knowledge of
art led to stronger preferences. In Table 1, we list unadjusted means. those who thought consciously or who did not think at all.
A. Dijksterhuis, Z. van Olden / Journal of Experimental Social Psychology 42 (2006) 627 631 631
The experiment extends earlier work by Dijksterhuis (2004; References
see also Dijksterhuis and Nordgren, in press) in two impor-
Betsch, T., Plessner, H., Schwieren, C., & Gütig, R. (2001). I like it but I
tant ways: Wrst, quality of decision was operationalized sub-
don t know why: A value-account approach to implicit attitude forma-
jectively rather than normatively, and second, participants
tion. Personality and Social Psychology Bulletin, 27, 242 253.
chose something real. Having people choose among hypo-
Bowers, K. S., Regehr, G., Balthazard, C., & Parker, K. (1990). Intuition in
thetical objects is one thing, but having people choose
the context of discovery. Cognitive Psychology, 22, 72 110.
among actual objects and then having their post-choice sat- Claxton, G. (1997). Hare brain, tortoise mind: How intelligence increases
when you think less. New York: HarperCollins.
isfaction aVected by how they arrived at their choices in the
Dijksterhuis, A. (2004). Think diVerent: The merits of unconscious
Wrst place is more intriguing. The current Wndings make our
thought in preference development and decision making. Journal of
earlier conclusion that unconscious thought can lead to
Personality and Social Psychology, 87, 586 598.
superior decisions much more ecologically valid and practi-
Dijksterhuis, A., & Meurs, T. (in press). Where creativity resides: The gen-
cally relevant. erative power of unconscious thought. Consciousness and Cognition.
Dijksterhuis, A., & Nordgren, L.F. (in press). A theory of unconscious
Our additional data on strength preferences and the rela-
thought. Perspectives on Psychological Science.
tion between the attitude towards the chosen poster and later
Dijksterhuis, A., Smith, P. K., van Baaren, R. B., & Wigboldus, D. H. J.
satisfaction shed light on why conscious thinkers sometimes
(2005). The unconscious consumer: EVects of environment on con-
make poor decisions. It is not the case that conscious thought
sumer behavior. Journal of Consumer Psychology, 15, 193 202.
leads to an absence of a strong preference. Instead, conscious Dijksterhuis, A., & de Vries, O. (2005). Unconscious thought is goal-
dependent. Unpublished data set, University of Amsterdam.
thought led to relatively strong preferences that later turned
Levine, G. M., Halberstadt, J. B., & Goldstone, R. L. (1996). Reasoning
out to be suboptimal. Wilson et al. (1993) have made the
and the weighing of attributes in attitude judgments. Journal of Person-
same argument when they refuted the possibility that con-
ality and Social Psychology, 70, 230 240.
scious thinkers were less conWdent in their preferences. Con-
Pelham, B. W., & Neter, E. (1995). The eVect of motivation on judgment
scious thought disturbs the decision process by letting people depends on the diYculty of the judgment. Journal of Personality and
Social Psychology, 68, 581 594.
put too much or too little weight on various attributes (see
Schooler, J. W., Ohlsson, S., & Brooks, K. (1993). Thoughts beyond words:
also Dijksterhuis & Nordgren, in press; Wilson & Schooler,
When language overshadows insight. Journal of Experimental Psychol-
1991) but it does not to an absence of a preference.
ogy: General, 122, 166 183.
The conclusion is that, within the conWnes of the current
Simonson, I., & Nowlis, S. M. (2000). The role of explanations and need
paradigm, unconscious thinkers did well. Imagine yourself for uniqueness in consumer decision making: Unconventional choices
based on reasons. Journal of Consumer Research, 27, 49 68.
at the auction in Paris again. The best strategy may be fol-
Wilson, T. D., Lisle, D., Schooler, J. W., Hodges, S. D., Klaaren, K. J.,
lowing: Wrst, take a good look at both the Monet and the
& LaFleur, S. J. (1993). Introspecting about reasons can reduce
Van Gogh. Then leave the auction and distract yourself for
post-choice satisfaction. Personality and Social Psychology Bulletin, 19,
a while (which is easy to do in Paris), and only then decide.
331 339.
Wilson, T. D., & Schooler, J. W. (1991). Thinking too much: Introspection
can reduce the quality of preferences and decisions. Journal of Person-
Acknowledgments
ality and Social Psychology, 60, 181 192.
Yaniv, I., & Meyer, D. E. (1987). Activation and metacognition of inacces-
We thank Teun Meurs for his help with the experiment
sible stored information: Potential bases for incubation eVects in prob-
and Jonathan Schooler for helpful comments on an earlier
lem-solving. Journal of Experimental Psychology: Learning, Memory,
draft. and Cognition, 13, 187 205.