Does unconscious thought outperform conscious thought on complex decisions? A further examination

Judgment and Decision Making, Vol. 4, No. 3, April 2009, pp. 235–247
Does unconscious thought outperform conscious thought on
complex decisions? A further examination
Todd J. Thorsteinson?
Scott Withrow
University of Idaho
Bowling Green State University
Abstract
Two experiments examined the bene?ts of unconscious thought on complex decisions (Dijksterhuis, 2004). Experi-
ment 1 attempted to replicate and extend past research by examining the effect of providing reasons prior to rating the
options. Results indicated no signi?cant differences between the conditions. Experiment 2 attempted to replicate the
?ndings of Dijksterhuis, Bos, Nordgren, and van Baaren (2006) and determine if a memory aid could overcome the
limitations of conscious thought on complex tasks. Results revealed that a memory aid improved decisions compared to
the conscious thought condition. Participants in the unconscious thought condition did not perform signi?cantly better
than did participants in the conscious thought condition.
Keywords: conscious thought, decision making, unconscious thought.
1 Introduction
1.1 Thinking too much
A growing body of literature has found evidence that
Conventional wisdom indicates that we should think
thinking too much can have detrimental effects on deci-
“hard” about our options when faced with a dif?cult
sion making and satisfaction with decisions (e.g., Halber-
decision. By devoting attention and conscious thought
stadt & Levine, 1999; Wilson & LaFleur, 1995; Wilson
to a dif?cult decision, one can carefully consider and
& Schooler, 1991). Wilson and Schooler (1991) exam-
weigh the various options and choose the option that
ined how thinking about the reasons for one’s decision
best matches one’s goals. Wilson and Schooler (1991)
may affect judgments. They proposed that thinking about
challenged this assumption when they found that partic-
how one feels about an object leads one to focus on plau-
ipants asked to think about their reasons for a decision
sible and salient factors, which may not be the real fac-
made apparently worse decisions than participants who
tors behind one’s feelings. For example, an individual
did not re?ect on their reasons for a decision. Recent
may justify his or her selection of a car based on the car’s
work (Dijksterhuis, 2004; Dijksterhuis, Bos, Nordgren,
high gas mileage, but the person’s preference for the car
& van Baaren, 2006) has cast further doubt on the ben-
may be due to the image the car conveys (e.g., the car is
e?ts of consciously thinking about one’s decisions by
“sporty”). It may be easier to justify, to oneself and oth-
demonstrating that consciously thinking about complex
ers, the purchase of a car based on gas mileage than on the
tasks (i.e., tasks which consist of numerous attributes that
image the car conveys. In addition, thinking about mul-
participants must weigh in order to determine the best op-
tiple attributes, as opposed to just a few attributes, may
tion) can lead to poorer performance than unconscious
highlight the fact that each alternative has some positive
thought. Some recent attempts to replicate this effect
and negative attributes and lead to more moderate evalu-
have been unsuccessful (Acker, 2008; Newell, Wong,
ations than if one had concentrated on a smaller subset of
Cheung, & Rakow, 2009; Rey, Goldstein, & Perruchet,
attributes (Linville, 1982).
2009). This paper attempts to replicate the bene?cial ef-
fects of unconscious thought on complex tasks found by
Wilson and Schooler (1991) provided evidence for the
Dijksterhuis and colleagues and to examine some condi-
detrimental effects of thinking about reasons on judg-
tions that may limit the effect.
ments. In Study 1, participants asked to think about
their reasons for liking or disliking different types of jam
showed less correspondence of their judgments with ex-
?We would like to thank Tanya Carr, Claudia Mahler, and Rylan
perts than did control participants. In Study 2, control
Clark for assistance with Study 1. Address all correspondence to Todd
participants were more likely to recall important infor-
Thorsteinson, Department of Psychology and Communication Studies,
P.O. Box 443043, University of Idaho, Moscow, ID 83844-3043. Email:
mation about courses and were more likely to preregister
tthorste@uidaho.edu.
for highly rated courses than participants asked to think
235

---

Judgment and Decision Making, Vol. 4, No. 3, April 2009
Unconscious versus conscious thought
236
about their feelings toward the courses. Thinking about
?rst three studies, which used similar methodology, par-
reasons for judgments or decisions may lead to judgments
ticipants were randomly assigned to one of three condi-
that depart from expert or consensual judgments.
tions, an immediate condition, a conscious thought con-
The apparent detrimental effects of thinking about rea-
dition, or an unconscious thought condition. Partici-
sons are not limited to subjective judgments about the
pants in the immediate condition made their judgments
taste of jam or characteristics that are important in college
immediately after viewing all the attributes of the tar-
courses. Halberstadt and Levine (1999) examined the ef-
get objects (i.e., apartments or roommates). Conscious
fect of thinking about reasons on predicting the outcomes
thought participants were asked to carefully think about
of basketball games. Participants who were asked to think
the target objects for a brief period (3 minutes in Ex-
about reasons for their judgments performed worse at
periments 1 & 2 and 4 minutes in Experiment 3). Un-
predicting winners of college basketball games than those
conscious thought participants were given a distracting
who were asked to make judgments based on intuition or
task to complete right after presentation of the attributes,
their “gut.”
for the same brief period as participants engaged in con-
These studies (e.g., Halberstadt & Levine, 1999; Wil-
scious thought (3 or 4 minutes). Although the task dif-
son & Schooler, 1991) documented possible harmful ef-
fered across the three experiments, each task was com-
fects of thinking about reasons for making judgments.
plex, involving the processing of a large amount of in-
However, thinking carefully about a judgment or decision
formation (e.g., 48 pieces of information in Experiment
does not require individuals to focus on reasons. Tordesil-
1). In each task, one of the target objects was created
las and Chaiken (1999) point out that there are different
to re?ect the best option (i.e., most of its attributes were
types of introspection and it is important to understand
positive) and one object re?ected the worst option (i.e.,
what types of introspection have negative effects on judg-
most of its attributes were negative). One or two ?ller ob-
ments. For example, Wilson and Schooler (1991, Study
jects was included in each experiment, which had an ap-
2) had two introspection conditions. In one condition,
proximately equal mix of positive and negative attributes.
the reasons condition, participants were asked to think
Participants in the unconscious thought condition showed
about the reasons for liking or disliking the various col-
greater differentiation between attractive and unattractive
lege courses. In the other condition, the “rate-all” condi-
target objects (i.e., greater difference in attractiveness rat-
tion, participants were asked to think about each piece of
ings between the attractive object and the unattractive ob-
information and rate how in?uential it was on their deci-
ject). In Experiments 4 and 5, Dijksterhuis (2004) exam-
sion. Although both the reasons condition and the rate-all
ined possible explanations for the results and discovered
condition had detrimental effects on judgment, the rate-
that unconscious-thought participants’ recognition of at-
all condition had a greater negative impact than the rea-
tributes appeared to be more polarized (i.e., greater recog-
sons condition.
nition of positive attributes of best roommate and neg-
ative attributes of worst roommate) and showed greater
clustering in memory (i.e., similar traits were more likely
1.2 Conscious and unconscious thought
to be recalled in order).
Although previous research has suggested that certain
Dijksterhuis and Nordgren (2006) proposed a theory
types of introspection can have detrimental effects (e.g.,
of unconscious thought that speci?ed why conscious
thinking about reasons), Dijksterhuis (2004; Dijksterhuis
thought may impair decision making when faced with
et al., 2006) has found that conscious thought performs
a complex decision. According to their unconscious-
worse than unconscious thought for complex decisions.
thought theory, conscious thought performs worse on
Conscious thought refers to “cognitive and/or affective
complex decisions because it has limited capacity, sub-
task-relevant processes one is consciously aware of while
optimal weighting of attributes, and engages in top-
attending to a task,” whereas unconscious thought refers
down processing that relies on schemata and expectan-
to “cognitive and/or affective task-relevant processes that
cies. Conscious thought has limited capacity in that
take place outside conscious awareness” (Dijksterhuis,
we can only hold a small amount of information in our
2004, p. 586). The research by Dijksterhuis (2004; Di-
consciousness, whereas the unconscious is considerably
jksterhuis et al., 2006) suggests that thinking about one’s
larger. Conscious thought interferes with the correct
options, even when not explicitly instructed to consider
weighting of attributes because it focuses on the most
reasons for one’s decision, can harm the quality of deci-
plausible and salient attributes, which are not always the
sions on complex tasks.
attributes that determine people’s judgments. Finally,
In the initial demonstration of the detrimental effects of
conscious thought relies more on schemata and expectan-
conscious thought, Dijksterhuis (2004) conducted a se-
cies because of the reduced demands on consciousness.
ries of studies examining whether unconscious thought
Conscious thought does perform well on simple tasks be-
outperforms conscious thought on complex tasks. In the
cause it follows rules and is precise.

---

Judgment and Decision Making, Vol. 4, No. 3, April 2009
Unconscious versus conscious thought
237
In order to directly test whether conscious thought per-
the highest expected value. Results indicated that the self-
forms better on simple tasks and unconscious thought
paced conscious thought condition outperformed both the
performs better on complex tasks, Dijksterhuis et al.
conscious thought condition and the unconscious thought
(2006) conducted a series of studies on what they called
condition in terms of choosing options with the highest
the “deliberation-without-attention” effect. In Study 1,
expected value. The ?ndings were consistent with their
conscious thought participants performed poorly on a
hypotheses that unconscious thought is not good at tak-
complex task but well on a simple task. Unconscious
ing into account magnitude and that conscious thought is
thought participants performed equally well on simple
detrimental when a ?xed time is required.
and complex tasks. In Study 2, they replicated the effects
Newell et al. (2009) also questioned whether un-
of Study 1 but this time with ratings of four cars (target
conscious thought would outperform conscious thought
objects) instead of a choice among them. In Study 3, they
when making decisions about options with attributes of
found a positive relationship between satisfaction with
varying degrees of importance. They contrasted two
purchases and amount of conscious processing for simple
rules, a weighted-additive model (WADD), where the va-
products and a negative relationship between satisfaction
lence of an attribute was weighted by its importance, with
with purchases and amount of conscious processing for
a simpler model where the option with the largest number
complex products. Finally, Study 4 was conducted in a
of positive attributes is preferred (TALLY). Dijksterhuis
naturalistic environment, where shoppers at a store with
and Nordgren (2006) proposed that unconscious thought
complex products were compared to shoppers at a store
is better than conscious thought at providing appropri-
with less complex products. For those shoppers at the
ate weights to attributes, but this was not tested in Di-
less complex product store, conscious thought and sat-
jksterhuis’s (2004; Dijksterhuis et al., 2006) studies as
isfaction with purchases was positively related, whereas
they used the TALLY model to de?ne the best option.
these measures were negatively related for those shoppers
Newell et al., in their ?rst experiment, created a deci-
at the complex products store. These studies provided
sion task involving four apartments similar to Dijkster-
strong evidence for the bene?ts of unconscious thought
huis’s (2004). They constructed the options so that the
for complex decisions.
apartment with the largest number of positive attributes
Other researchers, however, have had dif?culty repli-
was not the “best” apartment based on the weighting of
cating the bene?ts of unconscious thought over conscious
the attributes (WADD model; this apartment had fewer
thought for complex decisions. Some have argued that
positive attributes, but the positive attributes were more
further conscious processing may interfere with decision
important). The best apartment according to the WADD
making, so that it is not unconscious thought that is bene-
model was the most frequently chosen apartment for each
?cial, but that extended conscious processing is detrimen-
condition. Participants in the unconscious thought condi-
tal (Payne, Samper, Bettman, & Luce, 2008; Rey et al,
tion performed similarly to participants in the conscious
2009). Payne et al. (2008) proposed that requiring con-
thought condition. The similarity in performance sug-
scious processing for a ?xed time might be detrimental
gests that participants in the unconscious thought con-
because it provides too much time to think, causing deci-
dition were sensitive to differences in magnitude of the
sion makers to focus on attributes that are less important
apartments’ attributes.
to the decision. They included a conscious thought con-
The purpose of our research was to replicate the ?nd-
dition that was self-paced; that is, participants were able
ings of Dijksterhuis and colleagues regarding the bene?ts
to think consciously about the various options as long as
of unconscious thought over conscious thought on com-
they wished. They created two gambling tasks where
plex tasks and to explore the nature of the effect. Acker
participants selected a lottery from four choices. One
(2008) conducted a meta-analysis of studies comparing
choice maximized the chances of winning and another
conscious and unconscious thought and found a great
choice maximized the expected value of the gamble. In
deal of variability in effect sizes across studies. We in-
the ?rst gambling task, the option with the highest proba-
cluded a condition requiring participants to provide their
bility of winning and the option with the highest expected
reasons for their choice. This condition is similar to the
value had very similar expected values. Results indicated
conscious thought condition, but required participants to
that the self-paced conscious thought condition and the
write down their reasons for or against each target object.
unconscious thought condition performed equally well
Research has documented the detrimental effects of ask-
in terms of choosing options with high expected value,
ing participants to provide reasons for their choices (Hal-
and both conditions outperformed the conscious thought
berstadt & Levine, 1999; Wilson & Schooler, 1991) and
condition. In the second gambling task, they included
we were interested in comparing the reasons condition
a tradeoff between the probability of winning and maxi-
to a conscious thought condition. A reasons condition
mizing the expected value of the gamble, so that the op-
would ensure that participants were engaging in consid-
tion with the highest probability of winning did not have
erable thought about the options, and might further im-

---

Judgment and Decision Making, Vol. 4, No. 3, April 2009
Unconscious versus conscious thought
238
pair decision making relative to an unconscious thought
tion, but were told that each piece of information would
condition. Providing reasons may lead participants to fo-
be presented for four seconds. After participants read the
cus on more information, including information that is
instructions, 15 pieces of information about each apart-
not important for distinguishing among options.
ment were presented one at a time for four seconds each.
We also included a recall measure to help understand
Apartments and apartment attributes were presented in
the ?ndings. Tordesillas and Chaiken (1999) found that
an order randomized for each participant; however, all
participants asked to introspect about their decisions had
attributes of one apartment were presented before infor-
poorer recall of important information. We asked whether
mation about the next apartment was presented. Three
a similar effect might occur with conscious processing.
attributes for each apartment were included that were
Participants instructed to think carefully about a decision
not helpful in discriminating between the apartments, as
(conscious thought condition and reasons condition) may
the three attributes were either neutral in valence (e.g.,
spend too much time on information that is unimportant
type of trees near the apartment) or constant across all
to the decision.
apartments (e.g., “washers and dryers are available in the
building” was included in the description of all four apart-
2 Experiment 1
ments). The most attractive apartment was characterized
by nine positive attributes, three negative attributes, and
2.1 Method
three neutral attributes. The least attractive apartment was
characterized by nine negative attributes, three positive
2.1.1 Participants and design
attributes, and three neutral attributes. Two ?ller apart-
ments were included, both of which consisted of six pos-
Students (N = 153) were recruited from the psychology
subject pool at the University of Idaho and participated
itive attributes, six negative attributes, and three neutral
in exchange for course credit. The sample was relatively
attributes. (See the Appendix for listing of apartment at-
evenly divided between men (n = 71) and women (n =
tributes).
82). We asked participants to make attractiveness judg-
After the presentation of apartment attributes, the par-
ments about four apartments and to recall as many at-
ticipants in the immediate judgment condition rated the
tributes as possible. We counterbalanced the order so
attractiveness of the four apartments. Order of the attrac-
that half the participants recalled the apartment attributes
tiveness judgments was ?xed (i.e., all participants rated
prior to judging the attractiveness of the attributes and
Apartment A ?rst, then Apartment B, etc.). Participants
half judged the attractiveness of the apartments and then
in the conscious thought condition deliberated for four
recalled the apartment attributes.
minutes about which apartment they liked the best. In the
Participants were randomly assigned to conditions in
reasons condition, participants spent four minutes listing
a 2 (order: judgment before recall or recall before judg-
their reasons for liking and disliking the four apartments.
ment) x 4 (mode of thought: immediate judgment, con-
They were told the purpose was to organize their thoughts
scious thought, reasons, and unconscious thought) x 4
about which apartment they liked the best and that their
(apartments) mixed design, with the last factor as within-
responses would not be collected. Finally, participants in
subjects.
the unconscious thought condition worked on an n-back
Manipulating the order of recall and attractiveness
test (n-back of 2; Jonides et al., 1997) for four minutes as
judgments poses a problem for the unconscious thought
a distractor task.
condition.
Having participants in the unconscious
thought condition recall information before making judg-
Participants recalled as many attributes as possible
ments is likely to disrupt unconscious thought and make
about each apartment. Order of responses to the recall
the condition more similar to conscious thought. Our de-
task was randomized for each participant (i.e., some par-
sign allows us to test for this possibility and so we exam-
ticipants recalled attributes of Apartment C ?rst, some
ined possible interactions with order on all our dependent
recalled attributes of Apartment D ?rst, etc.). Prior to the
measures.
recall task, participants were not aware that they would
be asked to recall the apartment attributes. Half of the
participants completed the recall task prior to rating the
2.1.2 Procedure and stimulus materials
attractiveness of the apartments and the remaining half
Students participated in groups of up to ?ve individuals.
completed the recall task after rating the attractiveness
Stimulus materials were presented via computer using
of the apartments. At the end of the study, participants
MediaLab software (Jarvis, 2006). Participants evaluated
completed items about their satisfaction with the way in
four apartments based on the information presented. Par-
which they made the decision, their perceptions of the
ticipants were not informed about the amount of informa-
dif?culty of the task, and some demographic items.

---

Judgment and Decision Making, Vol. 4, No. 3, April 2009
Unconscious versus conscious thought
239
2.1.3 Measures
the conscious thought condition. Details regarding these
analyses and the analyses involving perceptions of the
1. Attractiveness of apartments. Participants were asked
task are provided below.
to rate the attractiveness of each apartment on a 9-point
scale ranging from 1 (not at all attractive) to 9 (extremely
attractive). We considered the highest rated apartment to
2.2.1 Attractiveness of the apartments
be the most attractive one to the participant.
A 2 (order: recall ?rst or judgment ?rst) x 4 (mode of
2. Decision satisfaction. Six items were included to
thought: immediate, conscious, reasons, unconscious) x
assess participants’ satisfaction with their decision mak-
4 (apartment) mixed ANOVA was conducted to deter-
ing process (e.g., “I am satis?ed with the way I reached
mine the effects of unconscious thought on ratings of
the judgments regarding the attractiveness of the apart-
the attractiveness of the apartments.1 Order and mode
ments,” “The process I used to judge the attractiveness
of thought were between-subject factors and apartment
of the apartments was effective”). These items were on
was a within-subject factor. We found no main effect for
a 9-point scale, ranging from 1 (strongly disagree) to 9
mode of thought, F(3,145) = 0.30, p = .82, partial ?2 =
(strongly agree). We removed one item as it lowered reli-
.01, and no main effect for order, F(1,145) = 0.86, p = .36,
ability. Coef?cient alpha for the ?ve items was 0.87.
partial ?2 = .01. None of the two-way or three-way inter-
3. Perceived ease of the task. Participants responded
actions was signi?cant (p’s > .30, partial ?2 ‘s < .01). The
to one item about the ease of the task (“The apartments
apartments were perceived as differing in attractiveness,
were easy to evaluate”), which was assessed on a 7-point
F(3,435) = 48.50, p < .001, partial ?2 = .25. As expected,
scale (1=strongly disagree to 7=strongly agree).
the best apartment was the most attractive (M = 7.04, SD
4. Recall of attributes. Participants listed as many at-
= 1.77) and the worst apartment was the least attractive
tributes as they could recall about the four apartments.
(M = 4.54, SD = 1.67). See Figure 1 for the attractiveness
For each apartment, we coded the total number of items
ratings for each apartment by mode of thought.
correctly recalled, number of positive items recalled,
The pattern of the means suggested that requiring par-
number of neutral items recalled, number of negative
ticipants to recall the attributes prior to making their judg-
items recalled, number of errors, number of positive er-
ments might have been harmful to the judgments made by
rors (i.e., attribute was negative but recalled as a positive
participants in the unconscious thought condition. The
attribute), and number of negative errors (i.e., attribute
three-way interaction was not signi?cant, but this may
was positive but recalled as a negative attribute). Both au-
have re?ected a lack of power, so we examined the two-
thors coded this information and were blind to the partic-
way interaction between order and apartment for each
ipants’ condition. Intraclass correlations, assessing con-
condition. None of the two-way interactions between or-
sistency and agreement, ranged from .89 to .99 (LeBreton
der and apartment approached signi?cance for the imme-
& Senter, 2008). Discrepancies between our ratings were
diate, conscious, and reasons conditions (p’s > .70, par-
resolved through discussion.
tial ?2 ‘s < .01). The order and apartment interaction ap-
peared to have some effect in the unconscious thought
condition, F(3,108) = 2.41, p = .07, partial ?2 = .06. As
2.2 Results and discussion
shown in Figure 2, the difference in attractiveness rat-
Ratings of the attractiveness of the apartments were sim-
ings between the best apartment and worst apartment was
ilar across mode of thought conditions. However, the or-
greater when judgment preceded recall than when recall
der of the recall and judgment tasks did appear to affect
preceded judgment. Asking participants to recall the at-
participants in the unconscious thought condition. When
tributes before making the judgments may have induced
judgments preceded recall, the difference in ratings be-
conscious thought in these participants. Easily recalled
tween the best and worst apartment was larger than when
attributes might have exerted a stronger in?uence on their
recall preceded judgments for the unconscious thought
judgments of attractiveness.
condition. Thus, when judgment preceded recall, our
1We conducted additional analyses to examine possible sex dif-
?ndings were consistent with those reported by Dijkster-
ferences. In Experiment 1 and Experiment 3 of Dijksterhuis (2004),
huis (2004; Dijksterhuis et al., 2006).
the difference in performance between the unconscious and conscious
Our analysis of the recall data tested whether partic-
thought conditions was found for men, but not for women. A 2 (sex:
male or female) X 4 (mode of thought: immediate, conscious, reasons,
ipants in the conscious thought condition would show
unconscious) X 4 (apartment) mixed ANOVA was conducted on attrac-
greater recall of neutral attributes compared to partici-
tiveness ratings to test for possible sex differences. We did not examine
pants in the unconscious thought condition. Our results
possible order effects because including that variable in the analyses
indicated no support for this hypothesis, as participants
would result in small sample sizes in each cell. There was no main
effect of sex on attractiveness ratings, F(1,145) = 1.18, p = .28, par-
in the unconscious thought condition recalled a higher
tial ?2 = .01, and no interaction between sex and mode of thought on
proportion of neutral attributes than did participants in
attractiveness ratings, F(3,145) = 0.38, p = .77, partial ?2 = .01.

---

Judgment and Decision Making, Vol. 4, No. 3, April 2009
Unconscious versus conscious thought
240
Unconscious
Judge.then.Recall
8
Reasons
8
Recall.then.Judge
Conscious
Immediate
6
6
4
4
Attractiveness Ratings
Attractiveness Ratings
2
2
Best
Worst
Avg1
Avg2
Best
Worst
Avg1
Avg2
Apartment
Apartment
Figure 1: Attractiveness ratings of the four apartments by
Figure 2: Attractiveness ratings in the unconscious
mode of thought. Error bars indicate one standard error.
thought condition. Error bars indicate one standard error.
Given that having the recall task precede the judgment
be some difference in proportion of neutral attributes re-
task may have disrupted unconscious thought, we con-
called as a function of mode of thought, F(3,99) = 1.99,
ducted a comparison between unconscious thought and
p = .12, partial ?2 = .06, but the pattern of means did not
conscious thought when judgment preceded recall. A
match our predictions. The unconscious thought (M =
2 (mode of thought: unconscious thought vs. conscious
0.22, SD = 0.19) and immediate conditions (M = 0.22, SD
thought) x 2 (apartment: best vs. worst) mixed ANOVA
= 0.16) recalled a higher proportion of neutral attributes
was conducted on attractiveness ratings. The interaction
than the conscious thought (M = 0.16, SD = 0.13) and
between mode of thought and apartment was not statis-
reasons conditions (M = 0.16, SD = 0.12), suggesting that
tically signi?cant, F(1,36) = 1.10, p = .30, partial ?2 =
neutral attributes did not receive undue focus from con-
.03, nor was the main effect for mode of thought, F(1,36)
scious thought participants.
= 1.60, p = .21, partial ?2 = .04, but the direction of
the means was consistent with Dijksterhuis’s ?ndings.
The pattern of means suggested an interaction between
The difference between the best and worst apartment was
mode of thought and apartment on proportion of neutral
greater for the unconscious thought condition (M = 3.56,
attributes recalled, F(9,297) = 1.63, p = .11, partial ?2
SD = 2.12) compared to the conscious thought condition
= .05. As shown in Figure 3, the conscious thought and
(M = 2.70, SD = 2.81; d = 0.34).
reasons condition had similar patterns of recall of neu-
tral attributes. Unconscious thought participants recalled
a higher portion of neutral attributes for the worst apart-
2.2.2 Recall of attributes
ment, which may have interfered with their ability to at-
Conscious thought might perform more poorly relative
tend to the more relevant attributes of the worst apart-
to unconscious thought if it focuses on less important at-
ment.
tributes. As there were differences in the number of at-
The interaction between order and mode of thought ap-
tributes recalled, we divided the number of neutral items
peared to affect the proportion of neutral attributes re-
recalled by the total number of items recalled for each
called, F(3,99) = 2.30, p = .08, partial ?2 = .07. As
apartment to get the proportion of neutral information re-
shown in Figure 4, all the mode of thought conditions, ex-
called. To examine whether conscious thought partici-
cept the unconscious thought condition, showed a higher
pants had better recall for neutral attributes compared to
proportion of recall of neutral attributes when judgment
unconscious thought participants, a 2 (order) x 4 (mode of
preceded recall. This pattern was reversed in the uncon-
thought: immediate, conscious, reasons, unconscious) x
scious thought condition. Recalling attributes prior to
4 (apartment) mixed ANOVA was conducted on the pro-
judging the options may have been especially disruptive
portion of neutral attributes recalled. There appeared to
to unconscious thought participants. Instead of relying

---

Judgment and Decision Making, Vol. 4, No. 3, April 2009
Unconscious versus conscious thought
241
0.4
0.4
Unconscious
Unconscious
Reasons
Reasons
Conscious
Conscious
Immediate
Immediate
0.3
0.3
0.2
0.2
Attractiveness Ratings
0.1
Proportion of Neutral Attributes
0.1
Best
Worst
Avg1
Avg2
Recall then Judge
Judge then Recall
Apartment
Order
Figure 3: Mean proportion of neutral attributes recalled
Figure 4: Mean proportion of neutral attributes recalled
by mode of thought and apartment. Error bars indicate
by order and mode of thought. Error bars indicate one
one standard error.
standard error.
on their holistic judgment obtained through unconscious
2.2.3 Decision satisfaction and perceived ease of task
processing, participants may have relied on their mem-
ories of attributes when completing the rating task. Di-
We expected that unconscious thought participants would
jksterhuis (2004; Experiment 2) found that participants
be less satis?ed with the decision process because they
in the unconscious thought condition were more likely to
were unable to devote signi?cant time to consider the
report making a global impression or judgment (55.6%)
apartments. This prediction was not supported, as there
than did participants in the conscious thought condition
were no differences between the mode of thought condi-
(26.5%). Having the recall task precede the judgment
tions in satisfaction, F(3,145) = 0.60, p = .62, partial ?2
task may have altered the decision strategy for partici-
= .01. The main effect for order (i.e., whether recall oc-
pants in the unconscious thought condition.
curred prior to or after judgments of the apartments) was
The analysis of the recall of neutral attributes revealed
not signi?cant, F(1,145) = 0.01, p = .94, partial ?2 = .00,
that participants in the conscious thought condition were
but the interaction between mode of thought and order
not overly focused on neutral attributes compared to par-
accounted for a small effect, F(3, 145) = 1.80, p = .15,
ticipants in the unconscious thought condition. Contrary
partial ?2 = .04. Follow-up analyses of the interaction re-
to expectations, participants in the unconscious thought
vealed that the effect of order on satisfaction with the de-
condition were not better at ignoring the neutral attributes
cision process was most pronounced in the unconscious
compared to conscious thought participants.2
thought condition. Unlike the other mode of thought con-
ditions, participants in the unconscious thought condi-
2We also examined differences in recall of positive and negative at-
tion were more satis?ed with the decision process when
tributes across modes of thought. Dijksterhuis (2004) reported evidence
judgment preceded recall than when recall preceded judg-
that unconscious thought participants recognized positive attributes of
ment, F(1,36) = 3.76, p = .06, partial ?2 = .10 (see Figure
the best option and negative attributes of the worst option more fre-
quently than negative attributes of the best option and positive attributes
5).
of the worst option. We examined whether a similar effect would occur
We also examined whether participants would differ in
with recall. We divided the number of positive attributes recalled for
their perceptions of the ease of the task. No signi?cant
each apartment and divided it by the total number of positive attributes
for each apartment, giving us a proportion of positive attributes recalled.
differences for the mode of thought conditions on percep-
Likewise, we divided the number of negative attributes recalled for each
apartment and divided it by the total number of negative attributes for
huis (2004) would be re?ected in a three-way interaction between mode
each apartment, giving us a proportion of negative attributes recalled.
of thought, apartment, and valence. The three way interaction was not
A 2 (order) x 4 (mode of thought) x 2 (apartment) x 2 (valence) mixed
signi?cant, F(3,124) = 0.05, p = .99, partial ?2 = .00, and an inspection
ANOVA was conducted, with the last two factors (i.e., apartment and
of the means indicated that the pattern was not consistent with Dijkster-
valence) as within-subject factors. Evidence consistent with Dijkster-
huis (2004).

---

Judgment and Decision Making, Vol. 4, No. 3, April 2009
Unconscious versus conscious thought
242
sions, even though it is the impressions of the apartments
that are driving the decision.
Unconscious
8
Reasons
Conscious
3 Experiment 2
Immediate
The differences between our stimulus materials and the
6
stimulus materials used by Dijksterhuis may explain our
inability to replicate his results. We conducted a second
4
experiment using the stimulus materials from Study 1 of
Dijksterhuis et al. (2006) to address this possibility. We
also sought to investigate whether the poor performance
Satisfaction with Decision Process
2
of conscious thought is due to memory limitations. Di-
jksterhuis and Nordgren (2006) proposed that a major
weakness of conscious thought for complex decisions is
its limited capacity. However, decision makers may uti-
Recall then Judge
Judge then Recall
lize aids to help overcome this limitation.
Two recent studies investigated whether memory lim-
Order
itations may explain the relatively poor performance of
conscious thought. Rey et al. (2009), utilizing the stim-
Figure 5: Satisfaction with the decision process by mode
ulus materials from Dijksterhuis et al. (2006), included
of thought. Error bars indicate one standard error.
a conscious thought condition where all 48 attributes
were written on a sheet of paper and provided to par-
tions of ease was found, F(3,145) = 0.36, p = .78, partial
ticipants as they deliberated for four minutes. Perfor-
?2 = .01. There was no effect due to order of recall and
mance in this condition was compared to an unconscious
judgment, F(1,145) = 0.02, p = .89, partial ?2 = .00, and
thought condition and an immediate condition. No statis-
no interaction effect between order and mode of thought,
tically signi?cant differences were found among the con-
F(3,145) = 0.90, p = .44, partial ?2 = .02. Participants
ditions, but participants in the immediate condition made
found the task to be neither overly dif?cult nor overly
the largest number of correct choices (80%), followed
easy, as the mean response was near the midpoint of the
by the unconscious thought condition (63.3%), and the
scale (M = 4.38, SD = 1.54).
conscious thought condition (50%). Unfortunately, their
study lacked a conscious thought condition where partic-
ipants relied on their memory for the attributes, making
2.2.4 Conclusions
it dif?cult to determine the bene?ts of providing all the
Although participants in the unconscious thought condi-
attribute information.
tions did not perform better on the task than conscious
Newell et al. (2009) addressed this issue in their sec-
thought participants, the pattern of means, when the re-
ond experiment. They included an immediate condition,
call task occurred after the ratings, was consistent with
an unconscious thought condition, a conscious thought
the ?ndings of Dijksterhuis (2004). We found no differ-
condition (where participants had to rely on their mem-
ence between the reasons condition and the other condi-
ory), and a conscious thought with information condition
tions, suggesting that asking participants to provide rea-
(where participants were provided with an information
sons for their decision does not interfere with decision
board displaying all the attributes for the options while
making more than asking participants to think about their
they deliberated). Memory dif?culties did not appear to
options. Thinking about reasons may be detrimental be-
be a problem as the conscious thought condition and the
cause it may focus decision makers on factors that are
conscious thought with information conditions chose the
easy to articulate and justify to others. Our failure to ?nd
best apartment at similar rates (82.6% and 78.3%, respec-
detrimental effects of the reasons condition compared to
tively). The conscious thought conditions chose the best
the other conditions may be because the task involved de-
apartment more frequently than did participants in the un-
scriptions of the options without experiencing them. Vis-
conscious thought condition (65.2%), but the difference
iting apartments may provide decision makers with im-
was not statistically signi?cant.
pressions about apartments that are in?uential in their ?-
We expanded on this research by investigating the ef-
nal choice, but may be dif?cult to articulate. Thinking
fects of a memory aid on conscious processing. Present-
about reasons may force decision makers to consider at-
ing all the attributes of the options overcomes memory
tributes that they can state clearly, as opposed to impres-
limitations, but it may disrupt conscious processing if it

---

Judgment and Decision Making, Vol. 4, No. 3, April 2009
Unconscious versus conscious thought
243
leads participants to focus on attributes that are relatively
had four minutes to think carefully about the cars before
unimportant to them. Payne et al. (2008) suggested that
they made a choice. Participants in the immediate deci-
thinking too much about a decision leads to a dilution
sion condition made their choice immediately following
effect, as less important information receives too much
the presentation of the car attributes.
attention. In Experiment 2, we examined whether the
After participants chose the car they liked the best, they
poor performance of conscious thought could be over-
rated the importance of each of the 12 attributes (e.g.,
come with a memory aid (i.e., asking participants to take
handling, gas mileage) on a scale of 1 (not at all impor-
notes during the presentation of the attributes). Not only
tant) to 7 (very important). After rating each attribute,
does this overcome memory limitations, but it also allows
participants answered some demographic questions and
participants to record attributes that they believe are the
were debriefed.
most relevant to their decision and leave off less impor-
tant attributes.
3.2 Results and discussion
3.1 Method
Our results failed to replicate Dijksterhuis et al.
(2006). The unconscious thought condition and con-
3.1.1 Participants and design
scious thought conditions chose the best car (Hatsdun)
We recruited participants (N = 162) from a psychology
at similar levels. Participants in the conscious thought
subject pool at the University of Idaho. Participants were
with notes condition chose the best car more frequently
randomly assigned to one of four conditions: an imme-
than the other conditions, but this difference was signi?-
diate decision condition, a conscious thought condition,
cant only for the comparison with the conscious thought
a conscious thought with notes condition, and an uncon-
condition.
scious thought condition. Seven participants were elimi-
We analyzed participants’ ratings of subjective impor-
nated from the data for not following directions (e.g., fail-
tance of the car attributes to determine if unconscious
ing to take notes) or responding randomly, resulting in a
thought participants made choices more consistent with
?nal sample of 155.
their importance ratings than other conditions. The un-
conscious thought condition did not perform signi?cantly
better than the other conditions.
3.1.2 Procedure and stimulus materials
The stimulus materials were from Dijksterhuis et al.
3.2.1 Car choice
(2006). Students participated in groups of up to ?ve in-
dividuals. Stimulus materials were presented via com-
The conscious thought with notes condition had the high-
puter using MediaLab software (Jarvis, 2006). Partici-
est percentage of correct decisions (i.e., chose the car,
pants chose the car they liked best from four cars. Forty-
Hatsdun, with the most positive attributes; 70.7%), fol-
eight attributes (twelve attributes for 4 cars) were pre-
lowed by the unconscious thought condition (56.8%), the
sented one at a time, in random order, for eight seconds.
immediate thought condition (52.5%), and the conscious
Each piece of information described a positive or negative
thought condition (48.6%; see Figure 6).
feature of the car. The “best” car (i.e., the Hatsdun) was
As demonstrated in Table 1, participants in the un-
categorized by 75% positive attributes and the “worst”
conscious thought condition did not perform signi?cantly
car (i.e., the Nabusi) was characterized by 25% positive
better than the other conditions. We did ?nd that the con-
attributes. Two additional cars, the Kaiwa and Dasuka,
scious thought with notes condition made signi?cantly
were included with 58% and 50% positive attributes, re-
more correct decisions than did participants in the con-
spectively (there was an error in the stimulus materials
scious thought condition.
listed in Dijksterhuis et al., 2006, so that the Kaiwa had
58% positive attributes as opposed to 50%).
3.2.2 Importance ratings
Participants in the conscious thought condition had
four minutes after the presentation of the information to
Consistent with Dijksterhuis (Dijksterhuis, 2004; Dijk-
think carefully about the cars. In the unconscious thought
sterhuis et al., 2006), we de?ned the most attractive car
condition, participants were distracted from thinking
as the one with the largest number of positive attributes.
about the cars for four minutes by working on a series
However, depending on which attributes were most im-
of anagrams. After four minutes had passed, participants
portant to an individual, a car with fewer positive at-
chose the car they liked the best. Participants in the con-
tributes may have been perceived as more attractive. Us-
scious thought with notes condition were asked to take
ing participants’ ratings of attribute importance, we ex-
notes on the cars and were told that the notes would not be
amined how well participants’ judgments matched up
collected. After the presentation of the information, they
with their subjective ratings of importance (similar to the

---

Judgment and Decision Making, Vol. 4, No. 3, April 2009
Unconscious versus conscious thought
244
Table 1: Results of comparisons between modes of
thought on choice.
100
Unconscious
Comparison
?2(1)
p
?
Conscious.with.Notes
Conscious
80
Immediate
Unconscious vs. immediate
0.14 .708 .04
Unconscious vs. conscious
0.49 .485 .08
60
Unconscious vs. conscious with notes 1.65 .199 .15
Conscious vs. conscious with notes
3.96 .047 .23
Choice
40
Table 2: Mean difference between maximum score and
20
score for chosen car by mode of thought.
Mode of thought
n
M
SD
Unconscious
37 10.54 19.15
Hatsdun
Kaiwa
Dasuka
Nabusi
Mean 57.15
Mean 28.3
Mean 12.5
Mean 2.025
Conscious with Notes 41 4.44
9.70
Car
Conscious
37 9.57 11.87
Immediate
40 7.55 10.27
Figure 6: Car choices by mode of thought.
analysis in Experiment 3 of Dijksterhuis, 2004). For each
male participants performed equally well in the conscious
of the four cars, we added the importance ratings for the
and unconscious conditions, but men performed signif-
positive attributes and subtracted the importance ratings
icantly better in the unconscious condition compared to
for the negative attributes. This provided a score for each
the conscious condition. We found that women chose
car for each participant. We calculated the difference be-
the best car more frequently (62.4%) than men (48.1%),
tween the maximum score and the score for the chosen
?2(1) = 2.91, p = .09, ? = .14. However, our results
car.3 Thus, if participants chose the car with the highest
are not consistent with Dijksterhuis, as male participants
score, the difference would be zero. The larger the differ-
in our study performed equally poorly in the conscious
ence score, the poorer the choice. As shown in Table 2,
thought condition and in the unconscious thought condi-
the conscious with notes condition had the smallest dif-
tion (see Figure 7). Male and female participants may
ference score, suggesting that they were choosing the car
have adopted different strategies for choosing among the
that best matched their subjective importance ratings. A
cars.
oneway ANOVA was conducted to test whether the dif-
ference scores were affected by mode of thought. No
overall difference was found, F(3,151) = 1.65, p = .18, ?2
4 General discussion
= .03, but linear contrasts were conducted to explore pos-
sible differences among the conditions. The conscious
Two experiments examined the bene?ts of unconscious
with notes condition had a smaller mean difference than
thought on complex decisions. Both experiments failed to
the mean difference for the other three conditions com-
?nd a signi?cant difference on performance between par-
bined, t(151) = 1.99, p = .05, d = .36. No signi?cant
ticipants in the conscious thought condition and those in
difference was found between the conscious and uncon-
the unconscious thought condition. Experiment 1 found
scious thought conditions, t(151) = 0.32, p = .75, d = .06.
that a reasons condition, where participants had to list
These results suggest that our inability to ?nd a bene?-
their reasons for their ratings, resulted in similar judg-
cial effect for unconscious thought is not due to subjec-
ments to the conscious thought condition. Experiment 2
tive preferences of the car attributes.
revealed that conscious thought combined with a memory
aid resulted in better judgments than conscious thought.
This highlights an important boundary condition for re-
3.2.3 Sex differences
search on unconscious thought. If decision makers take
We also checked for sex differences on performance, as
notes or review information, they can compensate for the
Dijksterhuis (2004; Experiments 1 and 3) found that fe-
limited capacity of conscious thought.
Although neither of our experiments found statistically
3We thank Jonathan Baron for suggesting this analysis.
signi?cant differences between performance in the con-

---