13. E. J. Eide, H. Kang, S. Crapo, M. Gallego, D. M. Virshup,
Methods Enzymol. 393, 408 (2005).
14. S. Martinek, S. Inonog, A. S. Manoukian, M. W. Young,
Cell 105, 769 (2001).
15. M. Abe, E. D. Herzog, G. D. Block, Neuroreport 11, 3261
16. C. Iitaka, K. Miyazaki, T. Akaike, N. Ishida, J. Biol. Chem.
280, 29397 (2005).
17. P. S. Klein, D. A. Melton, Proc. Natl. Acad. Sci. U.S.A. 93,
18. G. Adelmant, A. Begue, D. Stehelin, V. Laudet, Proc. Natl.
Acad. Sci. U.S.A. 93, 3553 (1996).
19. B. W. Doble, J. R. Woodgett, J. Cell Sci. 116, 1175
20. A. Balsalobre, F. Damiola, U. Schibler, Cell 93, 929 (1998).
21. F. Zhang, C. J. Phiel, L. Spece, N. Gurvich, P. S. Klein,
J. Biol. Chem. 278, 33067 (2003).
22. A. J. Gelenberg et al., J. Med. 321, 1489 (1989).
23. J. A. Ripperger, L. P. Shearman, S. M. Reppert,
U. Schibler, Genes Dev. 14, 679 (2000).
24. E. Iwahana et al., Eur. J. Neurosci. 19, 2281 (2004).
25. F. Benedetti et al., Neurosci. Lett. 368, 123 (2004).
26. R. H. Lenox, T. D. Gould, H. K. Manji, Am. J. Med. Genet.
114, 391 (2002).
27. M. Gottlicher et al. , EMBO J. 20, 6969 (2001).
28. C. J. Phiel et al., J. Biol. Chem. 276, 36734 (2001).
29. M. E. Dokucu, L. Yu, P. H. Taghert, Neuropsychopharma-
cology 30, 2216 (2005).
30. We thank the University of Pennsylvania Diabetes and
Endocrinology Research Center Vector Core (NIH DK
19525) and M. J. Birnbaum for GSK3b adenoviruses,
and J. D. Alvarez and A. Sehgal for helpful discussions.
Supported by NIH DK45586 (to M.A.L.) and NIH
MH058324 (to P.S.K.).
Supporting Online Material
Materials and Methods
Figs. S1 to S7
20 October 2005; accepted 14 January 2006
On Making the Right Choice: The
Maarten W. Bos, Loran F. Nordgren, Rick B. van Baaren
Contrary to conventional wisdom, it is not always advantageous to engage in thorough conscious
deliberation before choosing. On the basis of recent insights into the characteristics of conscious
and unconscious thought, we tested the hypothesis that simple choices (such as between different
towels or different sets of oven mitts) indeed produce better results after conscious thought, but
that choices in complex matters (such as between different houses or different cars) should be left
to unconscious thought. Named the ‘‘deliberation-without-attention’’ hypothesis, it was confirmed
in four studies on consumer choice, both in the laboratory as well as among actual shoppers, that
purchases of complex products were viewed more favorably when decisions had been made in the
absence of attentive deliberation.
ommon knowledge holds that thor-
ough conscious thought leads to good
decisions and satisfactory choices.
Whether purchasing a new car, a desktop com-
puter, or a pair of shoes, people generally
believe that serious conscious deliberation in-
creases the probability that they will make the
Bright[ choice. This idea applies especially to
choices between products that are complex,
multifaceted, and expensive. Whereas most
people are willing to buy a new set of towels
without much thought, they are unlikely to buy
a new car or outfit a new kitchen without
A second pervasive idea is that the quality
of a choice benefits from Bsleeping on it.[
Rather than (or in addition to) thinking con-
sciously, people usually feel that Bunconscious
thought[ is useful for making sound decisions.
Whereas conscious thought refers to thought or
deliberation while conscious attention is di-
rected at the problem at hand, unconscious
thought can be defined as thought or delibera-
tion in the absence of conscious attention di-
rected at the problem (1). An example of
unconscious thought is the following: One
compares two holiday destinations (say the
Costa Brava and T uscany) and does not know
what to decide. One puts the problem aside and
after 48 hours of not thinking about it con-
sciously, suddenly the thought BIt_s going to be
Tuscany![ pops into consciou sness. This thou ght
itself is conscious, but the transition from
indecision to a preference 2 days later is the
result of unconscious thought, or of deliberation
The scientific literature has emphasized the
benefits of conscious deliberation in decision
making for hundreds of years (2, 3). The idea
that conscious deliberation is the ideal (if not
always attainable) way to approach a d ecision
forms the backbone of classic (4, 5)aswellas
contemporary perspectives on decision making
(6, 7) and attitude formation (8, 9). In contrast,
the notion that unconscious thought is fruitful
hardly developed beyond the status of Bfolk
wisdom.[ It has been postulated or investigated
by scientists infrequently Ebut see (10–13)^.The
question addressed here is whether this view is
justified. We hypothesize that it is not.
First, conscious thought does not always
lead to sound choices. For example, participants
who chose their favorite p oster among a set of
five after thorough contemplation showed less
postchoice satisfaction than participants who
only looked at them briefly (14, 15). Further-
more, conscious deliberation can make multiple
evaluations of the same object less consistent
over time (16). Two reasons why conscious
deliberation sometimes leads to poor judgments
have been identified. First, consciousness has
a low capacity (17, 18), causing choosers to take
into account only a subset of the relevant
informationwhentheydecide(13, 19). Second,
conscious thought can lead to suboptimal weight-
ing of the importance of attributes (13–16):
We tend to inflate the importance of s ome at-
tributes at the expense of others, leading to
Conversely, unconscious thought, or thought
without attention, can lead to good choices
(13, 14). In a recent experiment, participants
read information about four apartments of dif-
ferent desirability (20). They were either asked
to choose their favorite immediately, or given
the opportunity to choose after a period of
conscious thought, or distracted for some time
Department of Psychology, University of Amsterdam,
Roetersstraat 15, 1018 WB, Amsterdam, the Netherlands.
*To whom correspondence should be addressed. E-mail:
Fig. 1. Percentage of participants who chose the
most desirable car as a function of complexity of
decision and of mode of thought (n 0 18 to 22 in
each condition). Error bars represent the stan-
dard error .
Fig. 2. Difference in attitude (on a scale of –25 to
þ25) toward the desirable and undesirable car as a
function of complexity of decision and of mode of
thought (n 0 12 to 14 in each condition). Error
bars represen t the standard error.
www.sciencemag.org SCIENCE VOL 311 17 FEBRUARY 2006
before they chose. In the third of these con-
ditions, participants could only engage in
unconscious deliberation: They knew they
would have to choo se later, but the distraction
task prevented them from devoting conscious
attention to the choice. Interestingly, un-
conscious thinkers made better decisions than
conscious thinkers or than immediate choosers
Recently, we formulated the Unconscious
Thought Theory (UTT) (21) about the strengths
and weaknesses of conscious thought and un-
conscious thought, that is, of deliberation with
and without attention. Two characteristics of
conscious and unconscious thought are impor-
tant in the current context. First, conscious
thought is rule-based and very precise (22, 23).
Unconscious thought can conform to rules in
that it detects recurring patterns, as the literature
on implicit learning shows (24). However, in
order to actively follow strict rules, conscious
attention is necessary. For example, one cannot
do arithmetic without conscious attention. This
capacity to follow rules makes conscious thought
more precise in decision making, because it
can strictly follow self-generated rules such as
not exceeding a maximum price. Second, as
alluded to earlier, conscious thought suffers
from the low capacity of consciousness, mak-
ing it less suitable for very complex issues.
Unconscious thought does not suffer from low
capacity. Indeed, it has been shown that during
unconscious thought, large amounts of infor-
mation can be integrated into an evaluative
summary judgment (13).
These characteristics of conscious and un-
conscious thought led us to postulate the
Bdeliberation-without-attention[ hypothesis, on
the relation between mode of thought or
deliberation (conscious versus unconscious)
and the complexity and quality of choice.
Complexity is defined as the amount of in-
formation a choice involves. A choice between
objects for which one or two attributes are
important (such as oven mitts or toothpaste) is
simple, whereas a choice between objects for
which many attributes are important (cars or
houses) is complex. Conscious thought is hy-
pothesized, due to its precision, to lead to good
choices in simple matters. However, because of
its low capacity, conscious thought leads to
progressively worse choices with more com-
plex issues. Unconscious th ought (i.e., deliber-
ation without attention) is expected, because of
its relative lack of precision, to lead to choices
of lower quality. However, the quality of choice
does not deteriorate with increased complexity,
allowing unconscious thought to lead to better
choices than conscious thought under com-
plex circumstances, this latter idea being the
kernel of the deliberation-without-attention
hypothesis. Quality of choice was operation-
alized both normatively (studies 1 and 2) as
well as subjectively (as postchoice satisfaction,
in studies 3 and 4).
Study 1. Participants were subjected to a 2
(mode of thought: conscious versus uncon-
scious) 2 (complexity of choice problem: sim-
ple versus complex) factorial design (25). All
participants read information about four hypo-
thetical cars. Depending on the condition, each
car was characterized by 4 attributes (simple) or
by 12 attributes (complex). The attributes were
either positive or negative. One car was char-
acterized by 75% positive attributes, two by
50% p ositive attributes, and one by 2 5% posi-
tive attributes (supporting online text). After
reading the information about the four cars,
participants were assigned either to a conscious
thought condition or to an unconscious thought
condition. In the conscious thought condition,
participants were asked to think about the cars
for 4 min before they chose their favorite car.
In the unconscious thought condition, partic-
ipants were distracted for 4 min (they solved
anagrams) and were told that after the period of
distraction they would be asked to choose the
The percentages of participants who chose
the best car are shown in Fig. 1. The crucial two-
way interaction supporting the deliberation-
without-attention hypothesis was significant
EF(1,76) 0 4.85, P G 0.04^. Unconscious think-
ers fared relatively well and showed no differ-
ences between conditions (F G 1, not significant).
Conscious thinkers generally made the proper
choice under simple conditions, but performed
poorly under complex circumstances EF(1,40) 0
4.95, P G 0.04^.
Study 2. For the second study we made one
change (25). Rather than asking for a choice,
we asked participants about their attitudes
toward each of the four cars. As the dependent
variable, we used the difference in attitude
toward the best car and the worst car. Again,
conscious thinkers were better able to differen-
tiate the quality of the cars under simple
conditions, whereas unconscious thinkers were
better able to differentiate the quality of the cars
under complex conditions E F(1,47) 0 5.63, P G
0.03^. The means are shown in Fig. 2.
Study 3. In a pilot study, undergraduate
students were asked how many aspects of a
product they would take into account in the
purchase of 40 different products. In this way,
we obtained an average Bcomplexity score[
for 40 different products (supporting online
For the actual study, other students were
presented with this list of 40 products. From the
list, they were asked to choose a product that
they had recently purchased and were asked the
following questions: Which product did you
purchase? Did you know the product before
you went on the shopping trip? How much did
you think about the product between seeing it
for the first time and b uying it? How satisfied
are you with the product?
To test our hypothesis, we distinguished
participants who thought (either consciously or
unconsciously) about their purchase from im-
pulse buyers who did not think much at all.
Hence, participa nts who indicated that they
bought a product they had never come across
before the shopping trip were not included,
leaving only participants who knew the product
beforehand (n 0 49).
It is impossible to know whether people are
engaged in unconscious thought by asking
them, so strictly speaking, we can only test
the relationship between conscious thought,
Fig. 3. The relation between mode of
thought and postchoice satisfaction
(on a scale of 1 to 7) for the six
products most frequently chosen in
study 3. Higher bars indicate more
satisfaction. The more complex the
product (on a scale of 1 to 5), the fur-
ther to the right it is shown. The
complexity score is given in parenthe-
ses. Participants were divided into
conscious and unconscious thinkers
on the basis of a median-split for
each product individually. Each bar
repres ents between two and five
Fig. 4. Postchoice satisfaction of IKEA (n 0 27)
and Bijenkorf (n 0 27) shoppers as a function
of mode of thought. Error bars represent the
17 FEBRUARY 2006 VOL 311 SCIENCE www.sciencemag.org
complexity, and quality. However, it follows
from our definition of conscious and uncon-
scious thought (according to which attention
to the problem at hand is the crucial distin-
guishing factor) that they are at least partly
dependent. At any one point in time, attention
is either directed at the decision under con-
sideration, or it is not; that is, at any partic-
ular point in time, either you are attending to
buying a car, or you are not. The more you
think about a decision consciously (that is,
with attention), the less time remains to think
about the same decision unconsciously (that
is, without attention).
We regressed the amount of thought and the
average number of aspects on postchoice satis-
faction. As expected, thinking does not make
people more satisfied, nor does complexity
(t_s G 1). However, the interaction of the two
parameters significantly predicted postchoice
satisfaction Et(48) 0 2.13, P G 0.04^.Correla-
tions were calculated between amount of
thought and postchoice satisfaction for three
categories of products: complex, medium, and
simple. For products of medium complexity, no
correlation was found Er(18) 0 –0.03^;for
simple products, a positive correlation was
found Er(15) 0 0.57, P G 0.03^; and for complex
products, a negative correlation was found
Er(16) 0 –0.56, P G 0.03^. As expected, the
more people thought consciously about simple
products, the more satisfied they were with
their purchase. Conversely, the more people
thought consciously about complex products,
the less satisfied they were with their purchase.
Figure 3 depicts satisfaction as a function of
mode of thought for the six most frequently
chosen products (26).
Study 4. On the basis of the pilot study to
study 3, two shops were selected: one where
people generally buy complex products (IKEA,
which sells mainly furniture) and one where
people generally buy simple products (Bijenkorf,
a department store like Macy_s that sells
clothes, clothing accessories, and kitchen ac-
cessories). At the exit, shoppers were asked the
following questions: What did you buy? How
expensive was it? Did you know the product
before you went on the shopping trip? and
How much did you think about the product
between seeing it for the first time and buying
it? A few weeks later, the shoppers were asked
(over the phone) how satisfied they were with
their purchases. As in study 3, participants who
indicated that they bought a product they had
never come across before the shopping trip
were not included.
We divided participants (Bthinkers[)onthe
basis of a median-split procedure into those
who engaged in much conscious thought
(conscious thinkers) and those who engaged
in little conscious thought (unconscious think-
ers). As expected, conscious thinkers reported
more p ostchoice satisfaction than uncon-
scious thinkers for Bijenkorf products (simple
products) EF(1,25) 0 6.52, P G 0.02^. The oppo-
site was true for the IKEA customers (complex
products), in which case unconscious thinkers
showed more postchoice satisfaction than con-
scious thinkers EF(1,25) 0 6.12, P G 0.02^
In sum, in four studies we demonstrated the
deliberation-without-attention effect. Conscious
thinkers were better able to make the best
choice among simple products, whereas un-
conscious thinkers were better able to make the
best choice among complex products. Among
people who knew the product they purchased
before they went on a shopping trip, the amount
of conscious thought was positively related to
postchoice satisfaction for simple products and
negatively related to postchoice satisfaction for
Our aim was to test the Bdeliberation-
without-attention[ hypothesis both in the labo-
ratory and among shoppers. In that sense, it is
important to view our set of studies as a whole
rather than as a series of individual studies.
Study 4 has unavoidable disadvantages such as
that the IKEA and Bijenkorf samples may have
differed (after all, different shops attract a
different clientele), which naturally opens the
potential for alternative explanations. There-
fore, study 3 was done in order to Bbridge[ the
laboratory studies with study 4. It has many of
the assets of study 4 (real choices between real
products with satisfaction as the dependent
variable), except that all participants were
Although we investigated choices among
consumer products in our studies, there is no a
priori reason to assume that the deliberation-
without-attention effect does not generalize to
other types of choices—political, managerial,
or otherwise. In such cases, it should benefit
the individual to think consciously about
simple matters and to delegate thinking about
more complex matters to the unconscious.
References and Notes
1. It is important to note that attention to the problem at
hand is the crucial distinction in our definitions of conscious
and unconscious thought. Thinking about buying a new
car while attention is directed at possible new cars is
conscious thought. Thinking about buying a new car
while attention is temporarily directed elsewhere is
unconscious thought. This distinction does not mean that
conscious thought only comprises conscious processes.
One can compare it to speech. Speech is a conscious
process (i.e., attention is directed at it while one speaks),
but it is in part dependent on accompanying unconscious
processes (such as processes responsible for syntax or
2. R. Descartes, The Philosophical Writings of Descartes
(Cambridge Univ. Press, Cambridge, 1984).
3. J. Locke, An Essay Concerning Human Understanding
(Penguin, London, 1689/1997).
4. I. L. Janis, L. Mann, Decision Making: A Psychological
Analysis of Conflict, Choice, and Commitment (Free
Press, New York, 1977).
5. H. A. Simon, Q. J. Econ. 69, 99 (1955).
6. J. R. Bettman, M. F. Luce, J. W. Payne, J. Consum. Res. 25,
7. D. Kahneman, Am. Psychol. 58, 697 (2003).
8. S. Chaiken, J. Pers. Soc. Psychol. 39, 752 (1980).
9. R. E. Petty, J. T. Cacioppo, in Advances in Experimental
Social Psychology, L. Berkowitz, Ed. (Academic Press,
New York, 1986), vol. 19, pp. 123–205.
10. As far as we know, the possibility of unconscious thought
(as well as the term ‘‘unconscious thought’’) was explicitly
used for the first time by Schopenhauer, who remarked
that ‘‘One might almost believe that half of our thinking
takes place unconsciously’’ (27).
11. G. Claxton, Hare Brain, Tortoise Mind: How Intelligence
Increases When You Think Less (Harper Collins, New York,
12. The idea of unconscious thought was also studied by
various researchers interested in incubation processes in
creativity. An example is K. S. Bowers et al.(28).
13. A. Dijksterhuis, J. Pers. Soc. Psychol. 87, 586 (2004).
14. A. Dijksterhuis, Z. van Olden, J. Exp. Soc. Psychol.,
15. T. D. Wilson et al., Pers. Soc. Psychol. Bull. 19, 331
16. G. M. Levine, J. B. Halberstadt, R. L. Goldstone, J. Pers.
Soc. Psychol. 70, 230 (1996).
17. G. A. Miller, Psych. Rev. 63, 81 (1956).
18. T. Nørretranders, The User Illusion: Cutting Consciousness
Down to Size (Viking, New York, 1998).
19. T. D. Wilson, J. W. Schooler, J. Pers. Soc. Psychol. 60, 181
20. Quality of decision was operationalized from a
normative perspective. One of the choice options was
made more desirable than the others because it
had been assigned more positive aspects than the
21. A. Dijksterhuis, L. F. Nordgren, Persp. Psychol. Sci.,
22. The important distinction between following rules and
merely conforming to them (and the need for conscious
attention in the former) was made by S. A. Sloman
23. E. R. Smith, J. DeCoster, Dual-Process Theories in
Social Psychology, S. Chaiken, Y. Trope, Eds. (Guilford,
New York, 1999), pp. 323–336.
24. Unconscious thought is reminiscent of implicit learning,
but there is an important difference. Implicit learning
refers to aspects of a task that are learned while
working on the task (and that are inaccessible to
consciousness). Unconscious thought refers to thought
processes that take place after the encoding of relevant
information. A good example of this definition of
unconscious thought is the groundbreaking work by
Stickgold and colleagues on learning during sleep.
See, e.g., (30, 31).
25. Materials and methods are available as supporting
material on Science Online.
26. We found a correlation between number of aspects and
amount of thought (r 0 0.54, P G 0.001): The more
complex a product is, the more people think consciously
when deciding to purchase it. Understandable as this
may be, our analysis suggests that people should do the
opposite, i.e., think unconsciously when deciding to
purchase a complex product. The correlation between
number of aspects and price was also significant
(r 0 0.45, P G 0.001): Expensive products were more
complex than inexpensive ones.
27. A. Schopenhauer, in Essays and Aphorisms (Penguin,
London, 1851/1970), p. 123.
28. K. S. Bowers, G. Regehr, C. Balthazard, K. Parker,
Cogn. Psychol. 22, 72 (1990).
29. S. A. Sloman, Psychol. Bull. 119, 3 (1996).
30. R. Stickgold, M. Walker, Trends Cogn. Sci. 8, 191 (2004).
31. R. Stickgold et al., Science 294, 1052 (2001).
32. We thank E. Neimeijer, L. Schreers, and R. Wassenberg
for help with conducting study 4. This research was
supported by a grant from Nederlandse Organisatie voor
Wetenschappelijk Onderzoek (016.025.030).
Supporting Online Material
Materials and Methods
20 October 2005; accepted 9 January 2006
www.sciencemag.org SCIENCE VOL 311 17 FEBRUARY 2006