Journal of Experimental Psychology: General

The Role of Mind-Wandering in Measurements of General

Aptitude
Michael D. Mrazek, Jonathan Smallwood, Michael S. Franklin, Jason M. Chin, Benjamin Baird,
and Jonathan W. Schooler
Online First Publication, April 2, 2012. doi: 10.1037/a0027968

CITATION
Mrazek, M. D., Smallwood, J., Franklin, M. S., Chin, J. M., Baird, B., & Schooler, J. W. (2012,
April 2). The Role of Mind-Wandering in Measurements of General Aptitude. Journal of
Experimental Psychology: General. Advance online publication. doi: 10.1037/a0027968
Journal of Experimental Psychology: General © 2012 American Psychological Association
2012, Vol. ●●, No. ●, 000 – 000 0096-3445/12/$12.00 DOI: 10.1037/a0027968

The Role of Mind-Wandering in Measurements of General Aptitude

Michael D. Mrazek Jonathan Smallwood

University of California, Santa Barbara Max Plank Institute for Human Cognitive and Brain Sciences,
Leipzig, Germany

Michael S. Franklin, Jason M. Chin, Benjamin Baird, and Jonathan W. Schooler

University of California, Santa Barbara

Tests of working memory capacity (WMC) and fluid intelligence (gF) are thought to capture variability
in a crucial cognitive capacity that is broadly predictive of success, yet pinpointing the exact nature of
this capacity is an area of ongoing controversy. We propose that mind-wandering is associated with
performance on tests of WMC and gF, thereby partially explaining both the reliable correlations between
these tests and their broad predictive utility. Existing evidence indicates that both WMC and gF are
correlated with performance on tasks of attention, yet more decisive evidence requires an assessment of
the role of attention and, in particular, mind-wandering during performance of these tests. Four studies
employing complementary methodological designs embedded thought sampling into tests of general
aptitude and determined that mind-wandering was consistently associated with worse performance on
these measures. Collectively, these studies implicate the capacity to avoid mind-wandering during
demanding tasks as a potentially important source of success on measures of general aptitude, while also
raising important questions about whether the previously documented relationship between WMC and
mind-wandering can be exclusively attributed to executive failures preceding mind-wandering (McVay
& Kane, 2010b).

Keywords: attention, mind-wandering, working memory capacity, fluid intelligence, executive function

The prospect of quantifying an individual’s general aptitude Detterman, 2004; Schmitt et al., 2009). However, the underlying
with a single variable has been a longstanding aspiration of both reasons for the predictive utility of measures of general aptitude
academic and commercial research. Motivated as much by prac- are still under investigation. We propose that mind-wandering—
tical concerns as scientific parsimony, measures of general apti- defined as a fluctuation of attention away from a task to unrelated
tude are frequently used when determining access to competitive concerns—is associated with impaired performance on these mea-
schools and employment. It is now well documented that perfor- sures of general aptitude, thereby partially explaining the reliable
mance in a wide variety of contexts can be predicted by measures correlations between such measures as well as their broad predic-
such as working memory capacity (WMC; Conway, Jarrold, Kane, tive utility.
Miyake, & Towse, 2007; Kane, Hambrick, & Conway, 2005), This proposal originates from a reassessment of the recently
fluid intelligence (gF; Deary, Strand, Smith, & Fernandes, 2007; demonstrated link between WMC and mind-wandering. Individu-
Rohde & Thompson, 2007; te Nijenhuis, van Vianen, & van der als with high WMC mind-wander less during daily life in circum-
Flier, 2007), and the Scholastic Aptitude Test (SAT; Frey & stances self-rated as challenging or effortful (Kane, Brown, et al.,
2007). Furthermore, high-WMC individuals report less mind-
wandering and make fewer errors during the sustained attention to
response task (McVay & Kane, 2009). These findings have moti-
vated a theoretical account proposing that mind-wandering is a
Michael D. Mrazek, Michael S. Franklin, Jason M. Chin, Benjamin
Baird, and Jonathan W. Schooler, Department of Psychological and Brain consequence of failures in executive control (McVay & Kane,
Sciences, University of California, Santa Barbara; Jonathan Smallwood, 2010b; see Smallwood, 2010, for a contrasting perspective). The
Department of Social Neuroscience, Max Plank Institute for Human Cog- executive failure account of mind-wandering builds on the view
nitive and Brain Sciences, Leipzig, Germany. that the complex span tasks used to measure working memory are
Michael D. Mrazek, Jonathan Smallwood, Michael S. Franklin, and indicative of an individual’s capacity for executive attention: the
Jonathan W. Schooler are supported through United States Department of ability to keep goal-relevant representations in a highly accessible
Education Grant R305A110277, awarded to Jonathan W. Schooler and state in the presence of distraction or interference (Engle & Kane,
Jonathan Smallwood. Benjamin Baird is supported by a National Science
2003; Kane, Brown, et al., 2007; Kane, Conway, Hambrick, &
Foundation Graduate Research Fellowship under Grant DGE-0707430.
Engle, 2007). Accordingly, complex span tasks measure executive
The content of this article does not necessarily reflect the position or policy
of the U.S. Government, and no official endorsement should be inferred. attention by requiring the recall of stimuli that are presented in
Correspondence concerning this article should be addressed to Michael alternation with a secondary task that intermittently draws atten-
D. Mrazek, Department of Psychological and Brain Sciences, University of tion away from the to-be-recalled items. Rather than merely testing
California, Santa Barbara, CA 93106. E-mail: mrazek@psych.ucsb.edu an individual’s capacity to stay focused, complex span tasks create

1
2 MRAZEK ET AL.

a mandatory distraction and examine one’s ability to remember Salthouse & Pink, 2008). On the basis of these findings, they
items despite it. suggested that attention may be a possible link between these two
On the assumption that complex span tasks measure execu- measures. Consistent with this proposal, lapses of sustained atten-
tive attention, the executive failure view of mind-wandering tion indexed by slow responses during a vigilance task are related
interprets a negative correlation between performance on com- to both WMC and gF (Unsworth, Redick, Lakey, & Young, 2010).
plex span tasks and mind-wandering in other contexts as evi- Additionally, sustained attention and working memory predict
dence that mind-wandering results from a failure of executive overlapping parts of gF (Schweizer & Moosbrugger, 2004). Al-
control. However, there is an alternative explanation for why though gF measures do not include an unrelated processing task
complex span tasks predict mind-wandering. While perfor- and are therefore less obvious measures of executive attention,2
mance during a complex span task clearly depends on one’s mind-wandering is an equally applicable source of potential dis-
ability to appropriately handle the distraction of the unrelated
traction in both types of tasks. Further clarification of the relation-
processing task (which serves to briefly divert attention away
ship between WMC and gF could therefore come from an inves-
from the to-be-remembered items), another simple yet unexam-
tigation of mind-wandering that occurs during the testing of these
ined possibility is that task-unrelated thoughts (TUTs) also
constructs.
disrupt performance on the working memory measures them-
selves. Indeed, mind-wandering is a ubiquitous phenomenon
associated with reduced awareness of task stimuli and the external Experimental Overview
environment (Barron, Riby, Greer, & Smallwood, 2011; Kam et
al., 2011; Smallwood, Beach, Schooler, & Handy, 2008), impaired Four studies examined the role of mind-wandering during the
vigilance (Cheyne, Solman, Carriere, & Smilek, 2009; McVay & measurement of general aptitude with the primary aims of
Kane, 2009; Smallwood et al., 2004, 2008), absent-minded forget- establishing whether (a) measures of WMC that have been used
ting (Smallwood, Baracaia, Lowe, & Obonsawin, 2003), deficits in to make theoretical claims about mind-wandering are them-
random number generation (Teasdale et al., 1995), and poor read- selves confounded by mind-wandering and (b) whether mind-
ing comprehension (Reichle, Reineberg, & Schooler, 2010;
wandering is correlated with WMC, gF, SAT scores, and the
Schooler, Reichle, & Halpern, 2004; Smallwood, 2011; Small-
shared variance between these tests. Study 1 measured mind-
wood, McSpadden, & Schooler, 2008). Given that mind-
wandering during three widely used WMC tasks with the hy-
wandering is associated with such a wide range of performance
pothesis that mind-wandering during the completion of each
deficits (for a review, see Schooler et al., 2011), it seems likely that
mind-wandering that occurs during complex span tasks would also measure would be associated with lower WMC. Study 2 exam-
be associated with disruptions to performance.1 This suggests that ined the trial-by-trial co-occurrence of mind-wandering and
the relationship between WMC and mind-wandering on an unre- impaired WMC performance. This served to establish the rela-
lated task (e.g., McVay & Kane, 2009) may result at least in part tionship between mind-wandering and WMC even within a
from the association between mind-wandering during complex given individual’s performance, while also ruling out a variety
span tasks and the resulting estimation of WMC. of third-variable explanations. To clarify the relationship be-
If estimations of WMC are determined not only by one’s ability tween mind-wandering and WMC within the context of an
to handle the external distraction of the unrelated processing task experimental paradigm, Study 3 examined whether financial
but also by the internal distraction of mind-wandering, this would incentives would reduce mind-wandering and thereby improve
have important implications for the types of evidence that have WMC performance. Finally, Study 4 measured mind-wandering
been used to support the executive failure view of mind- during tests of both WMC and gF with the hypotheses that
wandering. Linking WMC to mind-wandering in other contexts mind-wandering during these tasks would be associated with
does not indicate that mind-wandering is a result of executive SAT scores and with a latent variable capturing the shared
failure if estimations of WMC are not independent of mind- variance between multiple measures of general aptitude.
wandering. Nonetheless, even mind-wandering during complex
span tasks could in principle result from executive failures. Rather
than falsifying the executive failure view of mind-wandering,
evidence that complex span tasks are themselves influenced by 1
In principle, one could maintain or recover access to task-relevant
mind-wandering would indicate that new lines of evidence would representations despite the distraction of mind-wandering in much the
be necessary to determine the precise relationship between these same way that participants recall to-be-remembered items in complex span
constructs. tasks despite the distraction of the unrelated processing task. This indicates
If mind-wandering occurs during the measurement of WMC and mind-wandering cannot simply be defined as executive failure because the
is associated with worse performance, the same may be true of occurrence of mind-wandering does not guarantee that the representations
of to-be-remembered items will be lost. The occurrence of mind-wandering
other measures of general aptitude. This suggests that mind-
is only an executive failure if the task-relevant representations are lost as
wandering may contribute to the well-established correlation be- a result.
tween WMC and gF (e.g., Conway, Cowan, Bunting, Therriault, & 2
The term executive control is used in numerous ways to refer to a
Minkoff, 2002; Engle, Tuholski, Laughlin, & Conway, 1999). varied set of cognitive processes. Existing work indicates that a popular
Unsworth and Engle (2005) decomposed a measure of gF (Raven’s measure of gF—Raven’s Progressive Matrices— does recruit a variety of
Advanced Progressive Matrices) and found that the relation be- executive control processes, some of which overlap with the executive
tween WMC and gF was fairly consistent across items regardless attention that complex span tasks are designed to measure (e.g., resolving
of the level of difficulty, memory load, or rule type (see also proactive interference from prior trials; Unsworth & Engle, 2005).
 MIND-WANDERING IN MEASUREMENTS OF GENERAL APTITUDE 3

Study 1 wandering during the crucial moment of task processing. After

answering the thought probe, participants were instructed that they
Participants would begin a new trial. A mind-wandering score was computed
for each task by calculating the mean of the three thought-probe
One hundred fifteen undergraduates (35 male) from the Univer- responses. A composite mind-wandering variable for each partic-
sity of California, Santa Barbara (UCSB), participated in exchange ipant was computed as the mean of the three mind-wandering
for course credit (mean age ⫽ 18.98 years, SD ⫽ 0.99). scores.

Procedure
Results and Discussion
All participants completed automated versions of the operation
span task (OSPAN), reading span task (RSPAN), and symmetry To examine whether thought sampling affected performance, we
span task (SSPAN) in a counterbalanced order (Unsworth, Heitz, compared the OSPAN scores of the participants in this study to
Schrock, & Engle, 2005). These complex span tasks present to- those of 97 participants (six excluded based on criteria specified
be-remembered stimuli in alternation with an unrelated processing above) who completed an unrelated study during the same period
task (i.e., verifying the accuracy of an equation in the OSPAN, the of the academic year. Univariate analysis of variance (ANOVA)
meaningfulness of a sentence in the RSPAN, and the vertical revealed no difference in proportion of items recalled between
symmetry of an image in the SSPAN). In each of 15 OSPAN and those who completed the OSPAN with experience sampling (M ⫽
RSPAN trials, the to-be-remembered items were sets of three to .651, SD ⫽ .228) or without (M ⫽ .622, SD ⫽ .231), F(1, 197) ⫽
seven letters chosen from a pool of 12 and presented for 250 ms. 0.915, p ⫽ .340. This indicates that the experience sampling did
In each of 12 SSPAN trials, participants recalled the location of not significantly impact performance.
two to five red squares presented for 650 ms within a 4 ⫻ 4 matrix. We then addressed the role of mind-wandering during the per-
The sequence of set sizes was standardized for all participants. formance of three WMC tasks. Table 1 presents the means, stan-
At the end of each trial, participants selected the presented items in dard deviations, and correlations among measures of mind-
the serial order in which they appeared. Following standard pro- wandering and WMC. Significant correlations were found between
cedure for these WMC tasks (Conway et al., 2005), eight partici- the three span scores as well as between the three mind-wandering
pants with accuracy rates of less than 85% on the unrelated scores. As hypothesized, within each of the three span tasks, those
processing task (including errors caused by failing to respond individuals who mind-wandered more during testing had lower
within a response deadline based on latencies [M ⫹ 2.5 SDs] for WMC scores. The correlation between the composite WMC and
15 practice items) were excluded from the analysis. Span scores composite mind-wandering variables was ⫺.40 (p ⬍ .001). These
were calculated as the total number of items recalled in correct findings suggest that measures of WMC may predict mind-
serial order across all trials (Conway et al., 2005). Because the wandering in other contexts (e.g., McVay & Kane, 2009) at least
total number of trials and stimuli vary across complex span tasks, in part because mind-wandering during these measures is associ-
a WMC composite for each participant was computed as the ated with the resulting estimates of WMC.
z-score average (mean) of the three span scores. Both subjective and empirical accounts suggest that mind-
At unpredictable intervals during each span task, three trial wandering is a graded phenomenon. Accordingly, measuring
response screens were replaced with thought-sampling probes mind-wandering using a 1–5 scale may capture variance that is
which asked participants to indicate to what extent their attention missed when using a dichotomous measure of on-task versus
was either on-task or on task-unrelated concerns using a 1–5 off-task (Christoff, Gordon, Smallwood, Smith, & Schooler, 2009;
Likert-type scale (1 ⫽ completely on-task, 2 ⫽ mostly on-task, 3 ⫽ Franklin, Smallwood, & Schooler, 2011; Mrazek et al., 2011). Yet,
both on the task and unrelated concerns, 4 ⫽ mostly on unrelated given the common practice of using dichotomous measures of
concerns, 5 ⫽ completely on unrelated concerns). This thought- mind-wandering, we also examined whether our findings would
sampling procedure provides the opportunity to assess mind- persist after transforming our continuous measurement of mind-

Table 1
Correlations Among Mind-Wandering and Working Memory Capacity Measures

Variable 1 2 3 4 5 6

1. OSPAN —
2. RSPAN .485ⴱⴱⴱ —
3. SSPAN .429ⴱⴱⴱ .237ⴱ —
4. OSPAN TUT ⫺.269ⴱⴱ ⫺.179 ⫺.255ⴱⴱ —
5. RSPAN TUT ⫺.239ⴱ ⫺.359ⴱⴱⴱ ⫺.292ⴱⴱ .501ⴱⴱⴱ —
6. SSPAN TUT ⫺.196ⴱ ⫺.123 ⫺.335ⴱⴱⴱ .457ⴱⴱⴱ .552ⴱⴱⴱ —
M 40.39 34.87 16.22 1.54 1.77 1.61
SD 13.58 13.56 6.35 0.59 0.79 0.70

Note. N ⫽ 107. OSPAN ⫽ operation span task; RSPAN ⫽ reading span task; SSPAN ⫽ symmetry span task;
TUT ⫽ task-unrelated thought.
ⴱ
p ⬍ .05. ⴱⴱ p ⬍ .01. ⴱⴱⴱ p ⬍ .001.
4 MRAZEK ET AL.

wandering into a dichotomous variable of on-task (when respon- sampling within each set size, 60 OSPAN trials were divided
dents reported a 1, indicating they were fully on task) versus equally into set sizes of three, five, or seven letters. On 60% of
off-task (when respondents reported a 2–5, indicating they were to these trials, thought probes using the same Likert-type scale as in
some extent thinking of task-unrelated concerns). This transfor- Study 1 occurred after the response screen (but before receiving
mation was the most appropriate for this data (as opposed to an trial feedback), asking participants to indicate their focus through-
extreme split such as 1–2 vs. 4 –5) because (a) any degree of out the prior trial. In total, 12 thought probes occurred at each set
off-task thought can be considered mind-wandering and (b) most size. Span scores were calculated as in Study 1. One participant
participants reported being fully on task at the majority of thought was excluded from the analysis for accuracy of less than 85% on
probes, making this transformation the equivalent of a median the unrelated processing task of the OSPAN. Retrospective reports
split. Using this approach, performance on each of the complex of TUT and TRI were collected using the thinking and content
span tasks was negatively correlated with the number of mind- component of the Dundee State Stress Questionnaire (Matthews,
wandering episodes during that task (OSPAN, r ⫽ ⫺.200, p ⫽ Joyner, Gilliland, Huggins, & Falconer, 1999). Test anxiety was
.039; RSPAN, r ⫽ ⫺.285, p ⫽ .003; SSPAN, r ⫽ ⫺.362, p ⬍ measured using the Reactions to Tests (RTT) scale (Sarason,
.001). The composite complex span score was also negatively 1984).
correlated with the total number of mind-wandering reports across
all three tasks (r ⫽ ⫺.338, p ⬍ .001). Results and Discussion

Study 2 In a replication of Study 1, overall mind-wandering was asso-

ciated with overall WMC (r ⫽ ⫺.426, p ⬍ .001). We next
Although Study 1 demonstrated that individual differences in addressed the trial-by-trial relationship between these variables.
mind-wandering during testing are associated with lower esti- OSPAN data are frequently analyzed using an absolute scoring
mates of WMC, a more comprehensive understanding can be method in which participants receive credit for a trial only if they
derived from examining whether mind-wandering on a given recall all of the items (Conway et al., 2005). This allows trials to
trial is associated with impaired performance on that trial. be binned based on accuracy. A paired-samples t test revealed that
Although a trial-by-trial analysis still cannot definitively indi- participants mind-wandered more during inaccurate trials (M ⫽
cate directionality in the relationship between mind-wandering 2.44) than accurate trials (M ⫽ 1.76), t(65) ⫽ 7.48, p ⬍ .001.
and WMC, it can rule out a number of possible alternative Because only 21 participants had variation in both accuracy and
explanations. For example, if the relationship occurs because mind-wandering within each of the three set sizes, we chose to use
people who mind-wander more under any circumstance also a trial-by-trial analysis in which each trial was treated as an
have working memory performance detriments, then we would individual case to test for an interaction between trial accuracy and
not necessarily expect to observe the co-occurrence of mind- set size. Each case was associated with a participant’s unique
wandering and impaired trial performance even within a given identifier to control for differences that existed between partici-
individual. Another possible account for the relationship between pants. Following the statistical procedure described in Bland and
mind-wandering and performance is that participants who are Altman (1995), mind-wandering was treated as the continuous
having difficulty with the task may struggle to remain engaged. outcome variable, accuracy (categorical) and set size (continuous)
This view predicts that the relationship between mind-wandering were included as predictor variables, and participant was treated as
and performance should be limited to trials in which task perfor- a categorical factor using dummy variables (therefore having 66
mance is challenging. If mind-wandering and poor performance degrees of freedom). This model was significant, R2 ⫽ .468, F(1,
co-occur even on trials with small set sizes, then it is less tenable 2342) ⫽ 30.18, p ⬍ .001. Accurate trials were associated with less
that mind-wandering is a consequence of task performance. Fi- mind-wandering (␤ ⫽ ⫺.199, p ⬍ .001), whereas set size was not
nally, to examine the additional possibility that the link between (␤ ⫽ .016, p ⫽ .374). As illustrated in Figure 1, there was also an
mind-wandering and WMC may be a consequence of task frustra- interaction between accuracy and set size such that the impact of
tion or test anxiety, Study 2 included a retrospective measure of set size was only relevant for inaccurate trials (␤ ⫽ .081, p ⫽
task-related interference (TRI; e.g., thinking about how well you .012). This interaction suggests that regardless of set size, attention
are performing) and a measure of trait test anxiety. If WMC is must be relatively focused on the task to achieve complete trial
unrelated to these measures, then we can be reasonably confident accuracy. By contrast, mind-wandering was greater at small set
that the relationship between mind-wandering and WMC is not the sizes than at large set sizes for inaccurate trials. This could be
product of task frustration or test anxiety. because errors at large set sizes may also be a result of the inherent
difficulty of these longer trials. Nonetheless, t tests revealed that
Participants the simple slopes at conditional values corresponding to the three
set sizes were all significantly different from zero, indicating that
Sixty-seven undergraduates (35 male) from the UCSB partici- mind-wandering predicted performance at each set size (3: t ⫽
pated in exchange for course credit (mean age ⫽ 18.64 years, 9.65, p ⬍ .001; 5: t ⫽ 16.08, p ⬍ .001; 7: t ⫽ 4.81, p ⬍ .001).
SD ⫽ 1.08). Mind-wandering may predict performance above and beyond
the overall trial accuracy, so we examined the trial-by-trial rela-
Procedure tionship between mind-wandering and the proportion of items
recalled. A multiple regression model was created in which pro-
Participants completed a version of the OSPAN that was the portion of items recalled was treated as the continuous outcome
same as in Study 1 except as indicated below. To provide sufficient variable, mind-wandering and set size were included as continuous
 MIND-WANDERING IN MEASUREMENTS OF GENERAL APTITUDE 5

Figure 1. Interaction of set size and trial accuracy on mind-wandering. Accurate trials were those in which all
stimuli were recalled in the correct serial position.

predictor variables, and participant was treated as a categorical and participant was treated as a categorical factor using dummy
factor using dummy variables. This model was significant, R2 ⫽ variables. This overall model was significant, R2 ⫽ .367, F(1,
.381, F(1, 2411) ⫽ 20.89, p ⬍ .001, with both mind-wandering 2411) ⫽ 23.90, p ⬍ .001, but prior trial accuracy did not explain
and set size explaining a significant amount of unique variance in a significant amount of unique variance in subsequent mind-
WMC. Impaired trial performance was associated with greater wandering (␤ ⫽ ⫺.026, p ⫽ .130). A similar model using pro-
mind-wandering (␤ ⫽ ⫺.263, p ⬍ .001) and larger set sizes (␤ ⫽ portion of items recalled correctly instead of overall trial accuracy
⫺.394, p ⬍ .001). As illustrated in Figure 2, there was an inter- was also significant, R2 ⫽ .368, F(1, 2411) ⫽ 24.052, p ⬍ .001,
action between mind-wandering and set size such that the impact and there was a modest association between proportion of items
of mind-wandering on trial performance was greater at larger set recalled on the prior trial and subsequent mind-wandering (␤ ⫽
sizes (␤ ⫽ ⫺.089, p ⬍ .001). Also, t tests revealed that the simple ⫺.066, p ⫽ .006). The fact that performance on a trial is weakly
slopes at conditional values corresponding to the three set sizes associated with mind-wandering on the next trial would be ex-
were all significantly different from zero, once again indicating pected if there is any consistency in these variables across the task
that mind-wandering was associated with impaired performance at (e.g., mind-wandering on Trial 1 is associated with a greater
each set size (3: t ⫽ 6.40, p ⬍ .001; 5: t ⫽ 21.00, p ⬍ .001; 7: t ⫽ likelihood of mind-wandering on Trial 2). To disentangle these
12.14, p ⬍ .001). relationships, a multiple regression model was created in which
We next conducted a lag analysis to determine whether poor proportion of items recalled was treated as the continuous outcome
performance on a trial led to more mind-wandering on the subse- variable, both mind-wandering on that trial and mind-wandering
quent trial. A multiple regression model was created in which on the subsequent trial were treated as continuous predictor vari-
mind-wandering was treated as the continuous outcome variable, ables, and participant was treated as a categorical factor using
prior trial accuracy was treated as a categorical predictor variable, dummy variables. This overall model was significant, R2 ⫽ .270,

Figure 2. Interaction of mind-wandering and set size on WMC. On-task and off-task are calculated as the 25th
and 75th percentiles of mind-wandering reports (such that on-task is ⬃1 and off-task is ⬃3 on the 1–5 scale).
6 MRAZEK ET AL.

F(1, 2411) ⫽ 6.469, p ⬍ .001. Whereas mind-wandering during a Study 3

trial continued to be a strong negative predictor of trial perfor-
Although the within-subject correlations between mind-
mance (␤ ⫽ ⫺.401, p ⫽ .000), mind-wandering on the subsequent
wandering and estimates of WMC suggest that mind-wandering
trial became a positive predictor of trial performance (␤ ⫽ .126,
may disrupt performance, the correlational designs of Studies 1
p ⫽ .001). This may indicate that when taking into consideration
and 2 are limited in their ability to support causal claims. Study 3
the consistency of mind-wandering across trials, performing
aimed to provide stronger evidence regarding the relationship
poorly on a trial is associated with slightly reduced mind-
between these constructs by examining whether the effect of
wandering on the subsequent trial. This interpretation is consistent
financial incentives—an experimental manipulation known to im-
with the recent demonstration that correlated activity between a prove complex span task performance (Heitz, Schrock, Payne, &
default mode region (posterior parietal cortex, thought to underlie Engle, 2008)—is mediated by a reduction in mind-wandering.
off-task thought) and a task-positive region (dorsolateral prefrontal
cortex, thought to underlie task performance) is associated with
Participants and Procedures
worse current trial performance but enhanced performance on the
subsequent trial (Prado & Weissman, 2011). One hundred thirty undergraduates (44 male) from the UCSB
Finally, we examined whether WMC was predicted by test participated in this study (mean age ⫽ 18.87 years, SD ⫽ 1.00).
anxiety or retrospective reports of either TRI and TUT (see Table All participants completed the OSPAN task as described in Study
2). As predicted, retrospective TUT was correlated with both the 1, except as described below. At five unpredictable intervals
online measure of mind-wandering (r ⫽ .426, p ⬍ .001) and WMC during the 20-trial OSPAN, a thought-sampling probe occurred
(r ⫽ ⫺.329, p ⫽ .007). By contrast, WMC was not significantly using the same procedure described in Study 2.
related to either TRI (r ⫽ ⫺.191, p ⫽ .121) or any of the subscales Half of the participants were randomly assigned to an experi-
from the RTT measure of test anxiety (Tension: r ⫽ ⫺.086, p ⫽ mental condition in which they would receive a financial incentive
.489; Worry: r ⫽ ⫺.181, p ⫽ .143; TUT: r ⫽ ⫺.104, p ⫽ .401; based on their task performance. Following several practice trials,
Bodily Sensations: r ⫽ ⫺.067, p ⫽ .592). these participants read instructions describing that a typical pay-
Study 2 provides further evidence that mind-wandering during ment could be as much as $5.00 and that it would be determined
complex span tasks is associated with lower estimations of WMC, equally by both their accuracy on the math problems and their
while also ruling out a variety of possible explanations for the ability to recall the letters. Following completion of the OSPAN
relationship. There was a co-occurrence of poor trial performance task and before payment, all participants were asked whether they
had believed that the financial incentive was a genuine offer. Five
and mind-wandering that persisted across set sizes. Given that all
participants expressed skepticism, with several reporting they had
participants had the capacity to perform well on trials requiring the
been deceived in other experiments. These participants were
recall of only three letters, the association between mind-
dropped from analyses, along with two participants who were
wandering and WMC is not simply a result of mind-wandering on
excluded for accuracy of less than 85% on the unrelated processing
trials that exceed one’s capacity. WMC was also not related to test
task of the OSPAN. At the end of the experimental session, all
anxiety or TRI, indicating that the link between mind-wandering
participants were paid $5.00.
and WMC was not merely an artifact of worrying or analyzing
one’s performance. By contrast, retrospective reports of TUT were
correlated with both the experience sampling measure of mind-
Results and Discussion
wandering and overall estimates of WMC. Finally, poor trial Replicating Studies 1–2, there was a negative correlation be-
performance did not result in more mind-wandering on the subse- tween mind-wandering and OSPAN performance (r ⫽ ⫺.211, p ⫽
quent trial, providing further evidence against the explanation that .02). As found by Heitz et al. (2008), participants who received the
poor WMC performance causes mind-wandering. financial incentive recalled more letters during the OSPAN (M ⫽
69.61, SD ⫽ 2.26) than those who did not (M ⫽ 63.40, SD ⫽
2.21), F(1, 123) ⫽ 3.87, p ⫽ .05. There was also significantly less
mind-wandering among those who received the financial incen-
Table 2 tives (M ⫽ 1.55, SD ⫽ .09) compared to those who did not (M ⫽
Correlations Among Thought and Performance Measures 1.82, SD ⫽ .09), F(1, 123) ⫽ 4.63, p ⫽ .03. On the basis of the
Variable 1 2 3 4 5
directional effects from Study 2, we examined a mediational model
in which financial incentives reduce mind-wandering and thereby
1. OSPAN — improve performance. As illustrated in Figure 3, the effect of the
2. OSPAN TUT ⫺.426ⴱⴱⴱ — financial incentive on performance was significantly mediated by
3. Retrospective TUT ⫺.329ⴱⴱ .552ⴱⴱⴱ —
mind-wandering. When considered in light of Studies 1–2, these
4. Retrospective TRI ⫺.191 .186 .333ⴱⴱ —
5. Test anxiety (RTT) ⫺.137 .211 .192 .364ⴱⴱ — results provide converging support that mind-wandering may dis-
M 186.46 1.99 1.98 2.85 2.02 rupt WMC performance.
SD 52.99 0.71 0.72 0.64 0.55

Note. N ⫽ 66. OSPAN ⫽ operation span task; OSPAN TUT ⫽ mind- Study 4
wandering during OSPAN; Retrospective TUT ⫽ mind-wandering mea-
sured after OSPAN; Retrospective TRI ⫽ task-related interference mea- An executive attention view of WMC emphasizes the ability to
sured after OSPAN; RTT ⫽ Reaction to Tests scale (all subscales). maintain and recover access to task-relevant stimuli that are peri-
ⴱⴱ
p ⬍ .01. ⴱⴱⴱ p ⬍ .001. odically unattended. Complex span tasks assess this ability by
 MIND-WANDERING IN MEASUREMENTS OF GENERAL APTITUDE 7

Figure 3. Bootstrapping was used to calculate a confidence interval around the indirect effect. A 95%
confidence interval based on 5,000 resamples was [0.01, 4.05]. Zero falls outside this confidence interval,
indicating that the mediation effect was significant (p ⬍ .05).

presenting to-be-recalled items in alternation with a secondary paper-and-pencil test. A mind-wandering score was calculated for
processing task. However, Studies 1–3 suggest that TUTs may each task by taking the mean of the eight thought-probe responses.
serve as an additional source of distraction. Given that mind- A composite mind-wandering variable for each participant was
wandering could be a source of distraction during any test, TUTs computed as the mean of the mind-wandering scores in the
may also be associated with performance on tests of gF. Study 4 OSPAN and RPM.
therefore embedded thought sampling into tests of both WMC Following these tasks, participants referenced the registrar’s
(measured using the OSPAN) and gF (measured using Raven’s website to report the exact SAT scores they submitted when
Progressive Matrices [RPM]) with the hypotheses that mind- applying to college. A mean score was computed for 15 partici-
wandering would (a) be associated with worse task performance, pants who took the SAT multiple times. Eleven participants had
(b) predict performance on the SAT taken by participants 1–3 not taken the SAT. Eight participants were excluded from the
years before (thereby providing important ecological validity for analysis for accuracy of less than 85% on the unrelated processing
the relationship described in this article), and (c) be strongly task of the OSPAN.
associated with a latent variable capturing the shared variance
between these measures of general aptitude. Results and Discussion
Study 4 addressed the role of mind-wandering during the testing
Participants of WMC (via the OSPAN) and gF (via RPM). Table 3 presents the
One hundred thirty-nine undergraduates (46 male) from the means, standard deviations, skew, kurtosis, and correlations
UCSB participated in exchange for course credit (mean age ⫽ among measures. Significant correlations were found between the
18.75 years, SD ⫽ 1.04). WMC and gF scores, as well as between the mind-wandering
scores across these two tests. In a replication of Studies 1 and 2,
mind-wandering during the OSPAN task was negatively correlated
Procedure
with WMC. Similarly, those individuals who mind-wandered more
All participants completed the OSPAN and the final two sets of during RPM performed less well on this test of gF. As hypothe-
RPM in a counterbalanced order. The OSPAN was administered as sized, mind-wandering that occurred during laboratory testing was
described in Study 1. RPM is a culture-fair measure of abstract also predictive of SAT performance. The correlation between the
reasoning (Raven, 1938). The two most challenging sets (D and E) composite mind-wandering variable and SAT scores was ⫺.38
were used, each consisting of 12 questions presented in ascending (p ⬍ .001).
order of difficulty. Each question consists of a 3 ⫻ 3 matrix of Study 4 provided an additional opportunity to assess whether
geometric figures with the bottom right figure missing. Following participants mistakenly responded to thought probes based on their
one practice question, participants were given 20 min to answer as
many questions as possible by selecting from eight alternatives the
3
figure that completes the overall series of patterns. A final score Although Study 1 found that thought sampling did not influence
performance on the OSPAN, 39 separate participants completed the same
was computed as the total number of correct solutions.
RPM task either with or without thought sampling to confirm that there
At eight unpredictable intervals during RPM, a thought-
was also no issue of reactivity in RPM. A univariate ANOVA revealed no
sampling probe asked participants to indicate to what extent their difference in total number of correct solutions between those with thought
attention was either on task or on task-unrelated concerns using the sampling (M ⫽ 19.32, SD ⫽ .655) and those without (M ⫽ 19.80, SD ⫽
same procedure described in Study 1.3 Participants were alerted to .638), F(1, 37) ⫽ 0.281, p ⫽ .600. Only 19 participants had thought-
answer these questions on the computer by a beep. After answering sampling data, but we observed a nonsignificant negative correlation
the thought probe, participants were instructed to resume their between mind-wandering and task performance (r ⫽ ⫺.183, p ⫽ .45).
8 MRAZEK ET AL.

Table 3
Correlations Among Mind-Wandering and Performance Measures

Variable 1 2 3 4 5

1. Raven’s —
2. OSPAN .241ⴱⴱ —
3. Raven’s TUT ⫺.370ⴱⴱⴱ ⫺.272ⴱⴱ —
4. OSPAN TUT ⫺.241ⴱⴱ ⫺.277ⴱⴱ .558ⴱⴱⴱ —
5. SAT .471ⴱⴱⴱ .314ⴱⴱⴱ ⫺.355ⴱⴱⴱ ⫺.317ⴱⴱⴱ —
M 18.23 35.34 1.66 1.59 1,808.92
SD 3.03 12.08 0.60 0.68 181.01
Skew (SE) ⫺0.16 (.21) ⫺0.31 (.21) 1.04 (.21) 1.04 (.21) ⫺0.18 (.22)
Kurtosis (SE) ⫺0.58 (.42) ⫺0.16 (.42) 1.28 (.42) 0.54 (.42) ⫺0.30 (.44)

Note. N ⫽ 131, except N ⫽ 120 for SAT analyses. Raven’s ⫽ Raven’s Progressive Matrices; OSPAN ⫽
operation span task; Raven’s TUT ⫽ mind-wandering during Raven’s; OSPAN TUT ⫽ mind-wandering during
OSPAN; SAT ⫽ Scholastic Aptitude Test.
ⴱⴱ
p ⬍ .01. ⴱⴱⴱ p ⬍ .001.

assessment of their task performance (i.e., reporting that they had would be associated with a latent variable capturing the shared
mind-wandered simply because they had performed poorly rather variance between these measures of general aptitude. As illustrated
than because they were actually mind-wandering). If mind- in Figure 4, a model with two latent variables was created using
wandering rates predict variance in SAT scores above and beyond Amos statistical software. Data screening indicated that the skew
WMC and gF task performance, this would be further evidence and kurtosis of the mind-wandering variables were within accept-
that mind-wandering reports are not simply indications of partic- able standards (Kline, 2011) and that all participants had scores
ipants’ appraisal of their task performance. A simultaneous regres- within four standard deviations of the mean for each analyzed
sion model predicting SAT scores from WMC, gF, and composite variable. The mind-wandering latent variable consisted of the
mind-wandering revealed that the three predictors explained ap- mind-wandering scores during the OSPAN and RPM. The general
proximately 29% of the variance in SAT scores, R2 ⫽ .289, F(1,
aptitude latent variable consisted of WMC, gF, and SAT scores.
116) ⫽ 15.74, p ⬍ .001. Those who mind-wandered more during
The resulting model had an adequate participant-to-parameter ratio
the laboratory tasks scored lower on the SAT (␤ ⫽ ⫺.17, p ⫽ .05,
of 11:1 (Kline, 2011). Each of the measures loaded significantly on
sr2 ⫽ .03), even when controlling for WMC and gF. This rela-
their respective constructs. The path from mind-wandering to
tionship indicates that participants’ responses to thought-sampling
probes were not merely reflections of their assessments of their general aptitude was ␤ ⫽ ⫺.70, indicating that mind-wandering
task performance. predicted 49% of the variance in general aptitude. Several statis-
Having confirmed that fluctuations of attention during testing tical tests confirmed that the fit of the model to the data was good,
are associated with WMC, gF, and SAT scores, structural equation ␹2(4, N ⫽ 120) ⫽ 2.891, p ⫽ .58; Hoelter Index ⫽ 391; standard-
modeling was used to test the hypothesis that mind-wandering ized root-mean-square residual ⫽ .029; root-mean-square error of
approximation ⫽ .000, 90% confidence interval [.000, .120]; com-
parative fit index ⫽ 1.000.4
Study 4 confirms that mind-wandering during tests of either
WMC or gF is associated with lower estimates of an individual’s
capabilities as indexed by two widely used and broadly predictive
tests. Moreover, the mind-wandering that occurred during labora-
tory testing was also predictive of scores on the SAT. This indi-
cates that the association between mind-wandering and perfor-
mance generalizes to an important measure of educational success
taken by more than a million students each year. Finally, the role
of mind-wandering during tests of general aptitude also suggests
that the reliable correlations between these measures may be at
least partially explained by the amount of TUT that occurs during
Figure 4. N ⫽ 120. Structural equation modeling for general aptitude and testing. Indeed, structural equation modeling suggests that as much
mind-wandering during testing. The path connecting the two latent vari-
as 50% of what is shared across measures of general aptitude can
ables (circles) reflects the association between the constructs. The numbers
from the latent variables to the manifest variables (rectangles) indicate the be explained by mind-wandering.
loadings of each measure onto the latent variable. All error terms represent
unexplained variance (1 ⫺ R2). gF.tut ⫽ task-unrelated thought during
4
Raven’s Progressive Matrices; WMC.tut ⫽ task-unrelated thought during Moderate skew and kurtosis of the mind-wandering variables were
the operation span task (OSPAN); gF ⫽ fluid intelligence assessed via within acceptable standards (Kline, 2011), but we nonetheless confirmed
Raven’s Progressive Matrices; WMC ⫽ working memory capacity as- that nonparametric path estimates were comparable and that logarithmic
sessed via the OSPAN; SAT ⫽ Scholastic Aptitude Test scores. transformation of these variables led to equivalent fit statistics.
 MIND-WANDERING IN MEASUREMENTS OF GENERAL APTITUDE 9

General Discussion tionship implied by claims that executive failures lead to mind-
wandering or that mind-wandering disrupts task performance. De-
These studies converge in support of the proposal that mind- finitive demonstrations of causality are challenging, especially
wandering during testing is consistently associated with lower when the variables in question cannot be directly manipulated (as
estimates of general aptitude. Individuals who mind-wandered we believe is currently the case for both mind-wandering5 and
more during WMC and gF testing performed less well, and an executive processes). Yet, even without direct experimental ma-
individual’s WMC performance was worse on those trials during nipulations, evidence of covariation, temporal precedence, and
which mind-wandering occurred. Furthermore, the performance elimination of alternative explanations can converge in support of
enhancement among individuals offered a financial incentive was causal claims (Cook, Thomas, & Campbell, 1979). The present
mediated by a reduction in mind-wandering. Finally, mind- studies have demonstrated covariation between mind-wandering
wandering during tests of WMC and gF predicted scores on the and performance both between and within individuals, presented a
SAT, a high-stakes test that weighs heavily in undergraduate variety of arguments supportive of a temporal sequence in which
admissions decisions. In fact, nearly 50% of the shared variance mind-wandering precedes poor performance rather than vice versa,
between WMC, gF, and SAT scores was explained by mind- and ruled out alternative explanations such as mind-wandering
wandering. Future research will be necessary to determine whether being a consequence of poor performance, task frustration, or test
other cognitive capacities tested by these measures (e.g., abstract anxiety. Given the longstanding tendency to ascribe causality to
reasoning) have predictive utility even when controlling for the simple correlations between mind-wandering and performance, the
mind-wandering that occurs during testing. It may be that a sub- current article represents a considerably more rigorous, if not
stantial proportion of what makes tests of general aptitude suffi- complete, demonstration that mind-wandering may disrupt task
ciently general is that they create a demanding task context in performance.
which mind-wandering is highly disruptive. Although general aptitude has traditionally been regarded as
The present studies also raise important questions regarding the unchangeable, recent evidence indicates that intensive training
role of executive control in mind-wandering. Prior work has ar- on working memory tasks produces improvements that gener-
gued that mind-wandering results from executive failure based on alize to tests of gF (Jaeggi, Buschkuehl, Jonides, & Perrig,
several strands of evidence, with perhaps the most central being 2008). While the cause of this improvement is unknown, the
the association between WMC and mind-wandering (Kane, present studies raise the exciting possibility that performance on
Brown, et al., 2007; McVay & Kane, 2009, 2010a, 2010b). How- tests of general aptitude might be improved by methods that reduce
ever, the present work demonstrates that the measurement of mind-wandering during testing. Consistent with this proposal, Jha,
executive attention using complex span tasks is confounded with Stanley, Kiyonaga, Wong, and Gelfand (2010) recently demon-
mind-wandering. The claim that mind-wandering results from strated that decrements in WMC resulting from stressful prede-
executive failure can therefore not be made on the basis of the ployment military training can be offset by mindfulness exercises.
association between complex span tasks and mind-wandering in Furthermore, Tang et al. (2007) found that meditation training
other contexts. improved performance on RPM (although only marginally more
Nonetheless, the present work does not rule out the possibility than a control condition). The conceptual and empirical link be-
that mind-wandering results from executive failure. Even the tween mind-wandering and mindfulness (Mrazek, Smallwood, &
mind-wandering during complex span tasks could in principle Schooler, 2012) suggests that future research should investigate
result from executive failures. New lines of evidence that eliminate whether the impact of mindfulness training on tests of general
or control for the occurrence of mind-wandering while measuring aptitude is mediated by a reduction in mind-wandering. Given the
executive control will help determine the precise relationship apparent costs of mind-wandering, strategies for reducing its oc-
between these constructs. Disentangling the causal relationship currence during demanding tasks may significantly improve per-
between mind-wandering and executive processes will ultimately formance in a broad range of critically important situations.
provide important insight into our understanding of these phenom-
ena. For instance, executive processes may facilitate mind-
wandering in contexts when prioritizing TUTs over task focus is 5
While our ongoing efforts indicate that mindfulness training may be a
adaptive (Baars, 2010). Indeed, recent findings suggest that at least particularly effective technique for reducing mind-wandering, mindfulness
under some circumstances mind-wandering can be functional both may also influence a variety of additional cognitive processes (such as
by promoting future planning (Baird, Smallwood, & Schooler, metacognitive regulation) and therefore cannot be considered an unpol-
2011) and enhancing creative incubation (Baird et al., 2012). If luted experimental manipulation of mind-wandering.
distracting thoughts can lead to task impairment but also possess
some functionality, then an important executive process may be References
the successful prioritization of mind-wandering over other com-
peting goals. One source of individual variation in mind- Baars, B. J. (2010). Spontaneous repetitive thoughts can be adaptive:
wandering may therefore be the number and salience of ongoing Postscript on “mind wandering.” Psychological Bulletin, 136, 208.
Baird, B., Smallwood, J., Mrazek, M. D., Kam, J., Franklin, M. S., &
task-unrelated goals that require conscious reflection.
Schooler, J. W. (2012). Mind-wandering facilitates creative incubation.
Causal data are not necessary when demonstrating that mind- Manuscript submitted for publication.
wandering both confounds measurements of executive control and Baird, B., Smallwood, J., & Schooler, J. W. (2011). Back to the future:
is strongly associated with the shared variance between measures Autobiographical planning and the functionality of mind-wandering.
of general aptitude, but further consideration of what constitutes a Consciousness and Cognition, 20, 1604 –1611. doi:10.1016/
causal demonstration is warranted given the inherent causal rela- j.concog.2011.08.007
10 MRAZEK ET AL.

Barron, E., Riby, L. M., Greer, J., & Smallwood, J. (2011). Absorbed in sensory cortex. Journal of Cognitive Neuroscience, 23, 460 – 470. doi:
thought: The effect of mind wandering on the processing of relevant and 10.1162/jocn.2010.21443
irrelevant events. Psychological Science, 22, 596 – 601. doi:10.1177/ Kane, M. J., Brown, L. H., McVay, J. C., Silvial, P. J., Myin-Germeys, I.,
0956797611404083 & Kwapil, T. R. (2007). For whom the mind wanders, and when: An
Bland, J. M., & Altman, D. G. (1995). Statistics notes: Calculating corre- experience-sampling study of working memory and executive control in
lation coefficients with repeated observations: Part 1. Correlation within daily life. Psychological Science, 18, 614 – 621. doi:10.1111/j.1467-
subjects. British Medical Journal, 310, 446. doi:10.1136/ 9280.2007.01948.x
bmj.310.6977.446 Kane, M. J., Conway, A. R. A., Hambrick, D. Z., & Engle, R. W. (2007).
Cheyne, J. A., Solman, G. J. F., Carriere, J. S. A., & Smilek, D. (2009). Variation in working memory as variation in executive attention and
Anatomy of an error: A bidirectional state model of task engagement/ control. In A. R. A. Conway, C. Jarrold, M. J. Kane, A. Miyake, & J. N.
disengagement and attention-related errors. Cognition, 111, 98 –113. Towse (Eds.), Variation in working memory (pp. 21– 48). Oxford, Eng-
doi:10.1016/j.cognition.2008.12.009 land: Oxford University Press.
Christoff, K., Gordon, A. M., Smallwood, J., Smith, R., & Schooler, J. W. Kane, M. J., Hambrick, D. Z., & Conway, A. R. A. (2005). Working
(2009). Experience sampling during fMRI reveals default network and memory capacity and fluid intelligence are strongly related constructs:
executive system contributions to mind wandering. PNAS: Proceedings Comment on Ackerman, Beier, and Boyle (2005). Psychological Bulle-
of the National Academy of Sciences, USA, 106, 8719 – 8724. doi: tin, 131, 66 –71. doi:10.1037/0033-2909.131.1.66
10.1073/pnas.0900234106 Kline, R. B. (2011). Principles and practice of structural equation mod-
Conway, A. R. A., Cowan, N., Bunting, M. F., Therriault, D., & Minkoff, eling. New York, NY: Guilford Press.
S. (2002). A latent variable analysis of working memory capacity, short Matthews, G., Joyner, L., Gilliland, K., Huggins, J., & Falconer, S. (1999).
term memory capacity, processing speed, and general fluid intelligence. Validation of a comprehensive stress state questionnaire: Towards a
Intelligence, 30, 163–183. doi:10.1016/S0160-2896(01)00096-4 state big three? In I. Merville, I. J. Deary, F. DeFruyt, & F. Ostendorf
Conway, A. R. A., Jarrold, C., Kane, M. J., Miyake, A., & Towse, J. (Eds.), Personality psychology in Europe (pp. 335–350). Tilburg, the
(2007). Variation in working memory. New York, NY: Oxford Univer- Netherlands: Tilburg University Press.
sity Press. McVay, J. C., & Kane, M. J. (2009). Conducting the train of thought:
Working memory capacity, goal neglect, and mind wandering in an
Conway, A. R. A., Kane, M. J., Bunting, M. F., Hambrick, D. Z., Wilhelm,
executive-control task. Journal of Experimental Psychology: Learning,
O., & Engle, R. W. (2005). Working memory span tasks: A method-
Memory, and Cognition, 35, 196 –204. doi:10.1037/a0014104
ological review and user’s guide. Psychonomic Bulletin & Review, 12,
McVay, J. C., & Kane, M. J. (2010a). Adrift in the stream of conscious-
769 –786. doi:10.3758/BF03196772
ness: The effects of mind wandering on executive control and working
Cook, T. D., & Campbell, D. T. (1979). Quasi-experimentation: Design &
memory capacity. In A. Gruszka, G. Matthews, & B. Szymura (Eds.),
analysis issues for field settings. Chicago, IL: Rand McNally.
Handbook on individual differences in cognition (pp. 321–334). New
Deary, I. J., Strand, S., Smith, P., & Fernandes, C. (2007). Intelligence and
York, NY: Springer. doi:10.1007/978-1-4419-1210-7_19
educational achievement. Intelligence, 35, 13–21. doi:10.1016/
McVay, J. C., & Kane, M. J. (2010b). Does mind wandering reflect
j.intell.2006.02.001
executive function or executive failure? Comment on Smallwood and
Engle, R. W., & Kane, M. J. (2003). Executive attention, working memory
Schooler (2006) and Watkins (2008). Psychological Bulletin, 136, 188 –
capacity, and a two-factor theory of cognitive control. In B. H. Ross
197. doi:10.1037/a0018298
(Ed.), The psychology of learning and motivation: Advances in research
Mrazek, M. D., Chin, J. M., Schmader, T., Hartson, K. A., Smallwood, J.,
and theory (Vol. 44, pp. 145–199). New York, NY: Elsevier Science.
& Schooler, J. W. (2011). Threatened to distraction: Mind-wandering as
doi:10.1016/S0079-7421(03)44005-X
a consequence of stereotype threat. Journal of Experimental Social
Engle, R. W., Tuholski, S. W., Laughlin, J. E., & Conway, A. R. A. (1999). Psychology, 47, 1243–1248. doi:10.1016/j.jesp.2011.05.011
Working memory, short-term memory, and general fluid intelligence: A Mrazek, M. D., Smallwood, J., & Schooler, J. W. (2012). Mindfulness and
latent-variable approach. Journal of Experimental Psychology: General, mind-wandering: Finding convergence through opposing constructs.
128, 309 –331. doi:10.1037/0096-3445.128.3.309 Emotion. Advance online publication. doi:10.1037/a0026678
Franklin, M. S., Smallwood, J., & Schooler, J. (2011). Catching the mind Prado, J., & Weissman, D. H. (2011). Heightened interactions between a
in flight: Using behavioral indices to detect mindless reading in real key default-mode region and a key task-positive region are linked to
time. Psychonomic Bulletin & Review, 18, 992–997. doi:10.3758/ suboptimal current performance but enhanced future performance. Neu-
s13423-011-0109-6 roImage, 56, 2276 –2282. doi:10.1016/j.neuroimage.2011.03.048
Frey, M. C., & Detterman, D. K. (2004). Scholastic assessment or g? The Raven, J. (1938). Raven Progressive Matrices. Los Angeles, CA: Western
relationship between the Scholastic Assessment Test and general cog- Psychological Services.
nitive ability. Psychological Science, 15, 373–378. doi:10.1111/j.0956- Reichle, E. D., Reineberg, A. E., & Schooler, J. W. (2010). Eye movements
7976.2004.00687.x during mindless reading. Psychological Science, 21, 1300 –1310. doi:
Heitz, R. P., Schrock, J. C., Payne, T. W., & Engle, R. W. (2008). Effects 10.1177/0956797610378686
of incentive on working memory capacity: Behavioral and pupillometric Rohde, T. E., & Thompson, L. A. (2007). Predicting academic achieve-
data. Psychophysiology, 45, 119 –129. ment with cognitive ability. Intelligence, 35, 83–92. doi:10.1016/
Jaeggi, S. M., Buschkuehl, M., Jonides, J., & Perrig, W. J. (2008). Im- j.intell.2006.05.004
proving fluid intelligence with training on working memory. PNAS: Salthouse, T. A., & Pink, J. E. (2008). Why is working memory related to
Proceedings of the National Academy of Sciences, USA, 105, 6829 – fluid intelligence? Psychonomic Bulletin & Review, 15, 364 –371. doi:
6833. doi:10.1073/pnas.0801268105 10.3758/PBR.15.2.364
Jha, A. P., Stanley, E. A., Kiyonaga, A., Wong, L., & Gelfand, L. (2010). Sarason, I. G. (1984). Stress, anxiety, and cognitive interference: Reactions
Examining the protective effects of mindfulness training on working to tests. Journal of Personality and Social Psychology, 46, 929 –938.
memory capacity and affective experience in a military cohort. Emotion, doi:10.1037/0022-3514.46.4.929
10, 54 – 64. doi:10.1037/a0018438 Schmitt, N., Billington, A., Keeney, J., Oswald, F. L., Pleskac, T., Sinha,
Kam, J. W. Y., Dao, E., Farley, J., Fitzpatrick, K., Smallwood, J., Schooler, R., & Zorzie, M. (2009). Prediction of 4-year college student perfor-
J. W., & Handy, T. C. (2011). Slow fluctuations in attentional control of mance using cognitive and noncognitive predictors and the impact on
 MIND-WANDERING IN MEASUREMENTS OF GENERAL APTITUDE 11

demographic status of admitted students. Journal of Applied Psychology, Task unrelated thought whilst encoding information. Consciousness and
94, 1479 –1497. doi:10.1037/a0016810 Cognition, 12, 452– 484. doi:10.1016/S1053-8100(03)00018-7
Schooler, J. W., Reichle, E. D., & Halpern, D. V. (2004). Zoning-out Smallwood, J., McSpadden, M., & Schooler, J. W. (2008). When attention
during reading: Evidence for dissociations between experience and matters: The curious incident of the wandering mind. Memory & Cog-
meta-consciousness. In D. T. Levin (Ed.), Thinking and seeing: Visual nition, 36, 1144 –1150. doi:10.3758/MC.36.6.1144
metacognition in adults and children (pp. 204 –226). Cambridge, MA: Tang, Y.-Y., Ma, Y., Wang, J., Fan, Y., Feng, S., Lu, Q., . . . Posner, M. I.
MIT Press. (2007). Short-term meditation training improves attention and self-
Schooler, J. W., Smallwood, J., Christoff, K., Handy, T. C., Reichle, E. D., regulation. PNAS: Proceedings of the National Academy of Sciences,
& Sayette, M. A. (2011). Meta-awareness, perceptual decoupling and the USA, 104, 17152–17156. doi:10.1073/pnas.0707678104
wandering mind. Trends in Cognitive Sciences, 15, 319 –326. Teasdale, J. D., Dritschel, B. H., Taylor, M. J., Proctor, L., Lloyd, C. A.,
Schweizer, K., & Moosbrugger, H. (2004). Attention and working memory Nimmo-Smith, I., & Baddleley, A. D. (1995). Stimulus-independent-
thought depends on central executive resources. Memory & Cognition,
as predictors of intelligence, Intelligence, 32, 329 –347. doi:10.1016/
23, 551–559. doi:10.3758/BF03197257
j.intell.2004.06.006
te Nijenhuis, J., van Vianen, A. E., & van der Flier, H. (2007). Score gains
Smallwood, J. (2010). Why the global availability of mind wandering
on g-loaded tests: No g. Intelligence, 35, 283–300. doi:10.1016/
necessitates resource competition: Reply to McVay and Kane (2010).
j.intell.2006.07.006
Psychological Bulletin, 136, 2, 202–207. doi:10.1037/a0018673
Unsworth, N., & Engle, R. W. (2005). Working memory capacity and fluid
Smallwood, J. (2011). Mind wandering while reading: Attentional decoupling, abilities: Examining the correlation between operation span and Raven.
mindless reading and the cascade model of inattention. Language and Intelligence, 33, 67– 81. doi:10.1016/j.intell.2004.08.003
Linguistics Compass, 5, 63–77. doi:10.1111/j.1749-818X.2010.00263.x Unsworth, N., Heitz, R. P., Schrock, J. C., & Engle, R. W. (2005). An
Smallwood, J., Beach, E., Schooler, J. W., & Handy, T. (2008). Going automated version of the operation span task. Behavior Research Meth-
AWOL in the brain: Mind wandering reduces cortical analysis of exter- ods, 37, 498 –505. doi:10.3758/BF03192720
nal events. Journal of Cognitive Neuroscience, 20, 458 – 469. doi: Unsworth, N., Redick, T. S., Lakey, C. E., & Young, D. L. (2010). Lapses
10.1162/jocn.2008.20037 in sustained attention and their relation to executive control and fluid
Smallwood, J., Davies, J. B., Heim, D., Finnigan, F., Sudberry, M., abilities: An individual differences investigation. Intelligence, 38, 111–
O’Connor, R., & Obonsawin, M. (2004). Subjective experience and 122. doi:10.1016/j.intell.2009.08.002
attentional lapse: Task engagement and disengagement during sustained
attention. Consciousness and Cognition, 13, 657– 690. doi:10.1016/ Received August 19, 2011
j.concog.2004.06.003 Revision received December 23, 2011
Smallwood, J. M., Baracaia, S. F., Lowe, M., & Obonsawin, M. (2003). Accepted January 29, 2012 䡲