Consciousness and Cognition 27 (2014) 14–26

Contents lists available at ScienceDirect

Consciousness and Cognition

journal homepage: www.elsevier.com/locate/concog

On the link between mind wandering and task performance

over time
David R. Thomson ⇑, Paul Seli, Derek Besner, Daniel Smilek
Department of Psychology, University of Waterloo, Canada

a r t i c l e i n f o a b s t r a c t

Article history: Here we test the hypothesis that fluctuations in subjective reports of mind wandering over
Received 14 January 2014 time-on-task are associated with fluctuations in performance over time-on-task. In Study
Available online 26 April 2014 1, we employed a singleton search task and found that performance did not differ prior
to on- and off-task reports, nor did individual differences in mind wandering predict
Keywords: differences in performance (so-called standard analytic methods). Importantly however,
Attention we find that fluctuations in mind wandering over time are strongly associated with
Mind wandering
fluctuations in behavior. In Study 2, we provide a replication of the relation between mind
Performance
wandering and performance over time found in Study 1, using a Flanker interference task.
These data indicate (1) a tight coupling between mind wandering and performance over
time and (2) that a temporal-analytic approach can reveal effects of mind wandering
on performance in tasks where standard analyses fail to do so. The theoretical and
methodological implications of these findings are discussed.
Ó 2014 Elsevier Inc. All rights reserved.

1. Introduction

As most undergraduates would likely attest, the difficulty associated with paying attention, say, during a lecture, depends
on both the difficulty of the material being lectured on, but also, and perhaps more importantly, on the amount of time one is
required to attend. It has been argued that the longer we are required to perform a given task, the more likely it is that our
minds will wander to task-unrelated thoughts (Smallwood, 2010; Smallwood & Schooler, 2006). In addition, there has been
much work devoted to exploring the consequences of mind wandering on behavior in a wide range of tasks and contexts
such as reading (e.g., Feng, D’Mello, & Graesser, 2013; Thomson, Besner, & Smilek, 2013), driving (He, Becic, Lee, &
McCarley, 2011), and remembering (Riby, Smallwood, & Gunn, 2008; Thomson, Smilek, & Besner, in press), to name a
few. Surprisingly, however, with a few notable exceptions (Cunningham, Scerbo, & Freeman, 2000; McVay & Kane, 2012;
Smallwood et al., 2004), there is very little empirical evidence linking increases in mind wandering over time-on-task with
changes in performance over time-on-task. Understanding the relation between attention and performance over time holds
not only theoretical importance for the study of sustained attention and pedagogy, but is also of practical importance to
human factors researchers striving to mitigate performance decrements that are observed when humans monitor largely
automated systems for extended periods (so-called ‘vigilance tasks’, e.g. radar monitors, security screeners). The purpose
of the present work is to more closely examine the link between changes in mind wandering and performance over time.

⇑ Corresponding author. Address: Department of Psychology, University of Waterloo, 200 University Avenue West, Waterloo, Ontario N2L3G1, Canada.
Fax: +1 519 746 8631.
E-mail address: d5thomso@uwaterloo.ca (D.R. Thomson).

http://dx.doi.org/10.1016/j.concog.2014.04.001
1053-8100/Ó 2014 Elsevier Inc. All rights reserved.
 D.R. Thomson et al. / Consciousness and Cognition 27 (2014) 14–26 15

To date, examinations of mind wandering (defined as task-unrelated thought – TUT) over time-on-task have focused
almost exclusively on general group-level trends. For example, Teasdale et al. (1995) found that subjective reports of TUT
were more frequent in the second half of both perceptual-motor and memory tasks relative to the first half. In addition, prac-
tice with either task resulted in higher reports of TUT. Similarly, in a visual-discrimination task with low target probability
(i.e. a ‘vigilance’ task), Cunningham et al. (2000) found that over the course of the session, self-caught instances of TUT
increased in frequency. Importantly, it was also shown that as TUT increased over time-on-task, there was a commensurate
increase in response time (RT) and decrease in detection accuracy for target events. A similar relation was observed by
McVay and Kane (2012), who found that increases in TUT over time in a vigilance task (i.e., a sustained attention task in
which very few trials require a response) were accompanied by increases in both mean RT and RT variability on correct
detections. Likewise, it has been shown that instances of TUT are greater in the second half of the Sustained Attention to
Response Task (SART; Robertson, Manly, Andrade, Baddeley, & Yiend, 1997) relative to the first, while response errors (com-
mission errors) also increase from the first to second half of the task (Smallwood et al., 2004). And finally, in the context of a
classroom lecture, Risko, Anderson, Sarwal, Engelhardt, and Kingstone (2012) found that ‘off-task’ responses to thought
probes were greater in the second half of the lecture relative to the first, while memory for the presented material decreased
from the first half to the second. Taken together, the available evidence points to two consistent observations: (1) When one
is required to maintain the focus of attention on an external task for extended periods of time, episodes of mind wandering
(defined in the laboratory as ‘task-unrelated’ thought) become more frequent; and (2) performance seems to decline as a
function of time across a range of tasks.
The foregoing observations suggest that performance changes over time-on-task are tightly related to increases in mind
wandering over time. Indeed, it may be the case that our inability to maintain a high level of performance over time in many
task contexts may owe directly to an increase in the tendency to engage in mind wandering. Bolstering this claim is evidence
showing that overall rates of reported mind wandering predict performance in tasks such as the SART (Manly, Robertson,
Galloway, & Hawkins, 1999; Robertson et al., 1997), Metronome Response Task (MRT; Seli, Cheyne, & Smilek, 2013), and
semantic encoding (Thomson et al., in press). However, it has never been shown that changes in performance over time-
on-task have a direct relation to changes in mind wandering over time-on-task. As a result, it has been argued that instances
of off-task thought and performance decrements over time are largely independent (Head & Helton, in press) and that per-
formance declines owe to something else entirely, such as the depletion of information processing resources (Caggiano &
Parasuraman, 2004; Smit, Eling, & Coenen, 2004). In addition, there remains the possibility that any relation between per-
formance and mind wandering over time results from some other variable, such as a reduction in the efficiency of resource
distribution processes between on- and off-task thought over time (see Thomson et al., 2013). Clearly, more work is needed
to establish a tight coupling between performance decrements over time-on-task and increases in mind wandering over
time.
The primary goal of the present work was to more closely examine the correspondence between performance changes
and mind wandering changes as they occur over time. Specifically, we sought to examine this correspondence at the level
of individual traits, namely, examining whether individuals who show a greater increase in mind wandering over time-
on-task also concomitantly show a greater change in performance over time. In addition, we sought to examine whether
fluctuations in task-unrelated thought over a session within an individual predict fluctuations in performance over time
within the same individual.
Our secondary goal was a methodological one. Examining the direct relation between mind wandering and performance
over time-on-task has the added benefit of providing an additional analytic technique that might reveal an association
between mind wandering and performance even when other analytic methods (e.g., comparing performance prior to ‘on’
and ‘off’ task reports, or individual differences (correlational) methods) fail to do so. We examine these issues in two differ-
ent laboratory tasks, namely, a singleton search task and a Flanker interference task.

2. Study 1

In Study 1 we examined the relation between mind-wandering changes and performance changes over time-on-task in a
simple visual search task. We chose to use a singleton (i.e., a ‘pop-out’) search task because it is an easy task that is likely to
promote high rates of subjective reports of off-task thought (Forster & Lavie, 2009). We also expected the monotony and low
demand (in terms of attentional resources) of the task to promote changes in both mind wandering and (consequently) per-
formance over time-on-task. If so, we planned to make use of temporal changes in behavior to better elucidate the link
between mind wandering and performance.
The easy pop-out search task was also useful with respect to our secondary methodological goal. Pop-out search is known
to be much easier than conjunction search (Treisman & Gelade, 1980) and is argued to unfold in an entirely ‘bottom-up’ man-
ner (Theeuwes, 1991; Theeuwes, 1992 – but see Thomson & Milliken, 2012, 2013). Accordingly, we expected that the extre-
mely low demands on attentional resources would leave performance largely unaffected when participants report being ‘on-
task’ relative to when they report being ‘off-task’, or when overall rates of mind wandering are correlated with performance
at the individual difference level. However, we hoped to still observe a general relation between mind wandering and per-
formance when assessed as a function of time-on-task. That is, if both mind wandering rates and performance (response
time and accuracy) change over time-on-task, then there may be a direct relation between the two.
16 D.R. Thomson et al. / Consciousness and Cognition 27 (2014) 14–26

2.1. Method

2.1.1. Participants
Sixty undergraduates at the University of Waterloo (33 male, 27 female, mean age = 20.3 years) provided informed con-
sent and participated in exchange for course credit.

2.1.2. Stimuli and procedure

The experimental program was written in Python (www.python.org) and presented using PsychoPy software (Peirce,
2007). The experiment was run on a Mac Mini computer with a 2.40 GHz processor connected to a 24 in. Phillips 244E
LCD monitor, placed at an approximate viewing distance of 57 cm.
Participants were told that on each trial they would see five colored squares appear on the computer screen and that each
square would have a gap either on the left or right side. Importantly, one of the squares would be a different color from the
others (i.e. red amongst green or green amongst red). The task of the participant was to locate the odd-colored square (the
target) and indicate whether the gap was on the left side (by pressing the ‘z’ key) or the right side (by pressing the ‘m’ key).
Speed and accuracy were stressed. Additionally, participants were informed that at various points throughout the search
task, that they would be asked to respond to thought probes regarding the focus of their attention. Specifically, they were
provided with a written definition of mind wandering adapted from Smallwood, McSpadden, and Schooler (2007) which
read as follows: ‘‘During this experiment you will be asked at various points whether your attention is firmly directed
towards the task, or alternatively you may be aware of other things than just the task. Occasionally you may find as you
are performing the search task that you begin thinking about something completely unrelated to what you are doing; this
is what we refer to as ‘mind wandering’. These thought-probes will be signaled by instructions telling you which button to
press if your thoughts were on-task or off-task just prior to the probe.’’ Participants made an ‘on-task’ response by pressing
the ‘z’ key and an ‘off-task’ response by pressing the ‘m’ key.
Participants were first given the search task instructions, followed by four sample search trials. They were then provided
with the thought probe instructions as well as the definition of mind wandering, before viewing and responding to a sample
thought probe. Following this, they were given the opportunity to ask any questions of the experimenter, after which they
began the experiment. Participants completed 1000 search trials, of which half had ‘red’ targets and half had ‘green’ targets.
In addition, half of the trials required a ‘left’ response and half required a ‘right’ response. These trial types were randomly
intermixed. Each search trial began with the presentation of a central fixation cross that subtended 0.5° of visual angle and
was visible for 500 ms, after which the search trial was presented. Stimuli appeared in five of eight possible locations around
an invisible circle that subtended 11.5° of visual angle in diameter. Stimulus locations were determined randomly on each
trial. The search stimuli themselves (the squares) subtended 2.0° of visual angle and the gap subtended 0.5° of visual angle.
The search display remained on the screen until a response was made. Throughout the task, participants were required to
respond to 40 thought probes. These probes were presented pseudo-randomly, such that probes were separated by a min-
imum of 12 intervening search trials, and no probe occurred during the first 15 trials of the experiment. The session lasted
approximately 40 min.
In terms of analyses we assess the contribution of mind wandering (off-task responses to thought probes) to both
response time and accuracy. Response time (RT) is analyzed only for those trials in which a correct response was made.
In addition, we analyze data for RT and accuracy separately as a function of whether the target color on the current trial
is the same or different from the target color on the previous trial (repeat and switch trials respectively). We do this because
it is well-known that RTs are significantly faster on target repeat trials relative to switch trials (which is the ‘priming of pop-
out’ effect: Maljkovic & Nakayama, 1994). As a result, only trials following trials with correct responses are analyzed. First we
analyze the relation between performance and subjective reports of mind wandering over time-on-task (i.e. the ‘temporal
approach’) to address our primary theoretical goals. Second, we analyze performance as a function of thought probe response
(‘on’ versus ‘off’ task) as well as at the individual difference level. These latter analyses (which are very common in the mind
wandering literature) will be contrasted with our novel temporal analyses to address our secondary methodological goals.

2.2. Results

2.2.1. Temporal analyses

In the following analyses we examine subjective reports of mind wandering as a function of trial-block. A trial block is
defined as one-quarter of the total number of experimental trials (250). In order to examine the potential effects of mind
wandering on performance, we examine how mind wandering and performance measures (RT and accuracy) vary together
over time. We first examine mind wandering, RT, and accuracy separately as a function of block (see Fig. 1). An Analysis of
Variance (ANOVA) that treated Block (1, 2, 3, 4) as a within subject factor revealed a significant effect of Block on the pro-
portion of thought probes that received an ‘off-task’ report, F(3, 174) = 32.06, p < .001, g2p = .36. Moreover, there was a signif-
icant linear trend, F(1, 58) = 61.85, p < .001, g2p = .52, indicating that off-task reports increased in frequency as a function of time-
on-task. A closer inspection of Fig. 1a reveals that off-task reports level-off between blocks three and four, indicating that there
is likely a ceiling on the proportion of time participants will engage in off-task thought within the present task (in this case
about 60% of the time).
 D.R. Thomson et al. / Consciousness and Cognition 27 (2014) 14–26 17

Fig. 1. Proportion of ‘off-task’ reports (A), mean response times as a function of target color repeat/switch from the previous trial (B), and proportion correct
as a function of target color repeat/switch from the previous trial (C) in Experiment 1 are depicted as a function of time-on-task (block). Error bars represent
one standard error of the mean.

We conducted a similar ANOVA on RTs, but with both Block and Trial Type (target color repeat/switch) as within subject
factors. There was a significant main effect of Block on RT, F(3, 174) = 25.07, p < .001, g2p = .30, and a significant linear trend,
F(1, 58) = 41.18, p < .001, g2p = .42, indicating that RT decreased as a function of time-on-task. There was also a significant main
effect of Trial Type, F(1, 58) = 75.07, p < .001. g2p = .56; the priming of pop-out effect (Maljkovic & Nakayama, 1994). There was no
significant Block by Trial Type interaction, F(3, 174) = 2.50, p = .061, g2p = .04, however, there was a significant linear trend,
F(1, 58) = 4.24, p = .044, g2p = .07, indicating that the priming effect diminished over time.
The analysis of accuracy was identical to that of RT. There was a main effect of Block on accuracy, F(3, 174) = 19.26,
p < .001, g2p = .25, and a significant linear trend, F(1, 58) = 44.45, p < .001, g2p = .43, with accuracy decreasing as a function of
time-on-task. There was also a main effect of Trial Type, F(1, 58) = 7.27, p = .009, g2p = .11; the priming of pop-out effect. As in
the analysis of RT, there was no significant Block by Trial Type interaction (F < 1), and no linear trend, indicating that the prim-
ing effect did not change as a function of time-on-task.
The results of the temporal analyses conducted so far converge with those of prior studies in which both mind wandering
and performance are shown to change over time-on-task (i.e., Cunningham et al., 2000; McVay & Kane, 2012). In the tem-
poral analyses that follow, we attempt to directly assess the relation between mind wandering and performance changes
over time. First, in order to examine whether individuals with greater increases in mind wandering across block demon-
strated greater changes in performance across block, we calculated slopes representing changes in the proportion of off-task
reports as a function of block for each participant. We then compared these slopes to slopes representing changes in RT and
accuracy over experimental blocks for each participant, to assess whether increases in mind wandering predicted changes in
RT and accuracy at the individual subject level. Pearson-product moment correlations reveal that changes in mind wander-
ing over time-on-task significantly predict changes in accuracy, n = 59, r = .266, p = .041, but not changes in RT, n = 59,
r = .024, p = .857.
18 D.R. Thomson et al. / Consciousness and Cognition 27 (2014) 14–26

Next, in order to assess whether fluctuations in mind wandering across block predicted fluctuations in performance
(without regard to whether such changes increase or decrease in a linear fashion as was the case with the analysis of slopes
described above), we computed individual Pearson correlation coefficients for each participant. In other words, we correlate
mind-wandering rates across block with RT and accuracy across block for each participant. We then compute the mean cor-
relation coefficient for each series of correlations and compare this value to zero (which would indicate that overall, there is
no relation). We computed mean correlation coefficients between mind wandering and: target color repeat/switch RT, and
target color repeat/switch accuracy. One-sample t-tests revealed all of these means were all statistically different from zero,
t(56) = 3.27, p = .002, d = .44, t(56) = 4.34, p < .001, d = .58, t(56) = 4.08, p < .001, d = .55, and t(56) = 4.35, p < .001, d = .58,
respectively. These can be seen in Fig. 2. Overall, more mind wandering was associated with poorer accuracy and faster
RT within a given block.

2.2.2. Non-temporal analyses

Having demonstrated that mind wandering and performance are tightly linked over time (using novel analytic tech-
niques), here, we applied a commonly used analysis technique, assessing whether mind wandering had any effect on per-
formance (in our case RT and/or accuracy during the search tasks) by obtaining performance measures only from the five
trials prior to each thought probe for each participant (as per Bastian & Sackur, 2013; Dixon & Li, 2013; Kam et al., 2012;
Seli, Carriere, Levene, & Smilek, 2013; Seli et al., in press; Thomson et al., 2013; Uzzaman & Joordens, 2011; etc.). We then
computed means separately for trials prior to probes on which ‘on’ and ‘off’ task reports were provided. One participant
failed to provide any ‘on-task’ reports and so was not included in this analysis (yielding a sample size of 59). Mean RTs
and mean accuracy as a function of probe response and trial type is depicted in Fig. 3.
RTs were submitted to a repeated-measures ANOVA that treated Probe Response (on-task, off-task) and Trial Type (color
repetition, color switch) as within-subject factors. There was no main effect of Probe Response (F < 1), with RTs being similar
for the trials leading up to ‘on-task’ (M = 713 ms, SE = 13 ms) and ‘off-task’ (M = 720 ms, SE = 25 ms) reports. There was a sig-
nificant main effect of Trial Type, F(1, 57) = 82.48, p < .001, g2p = .59, with RTs being faster on target color repeat trials
(M = 660 ms, SE = 18 ms) than on switches (M = 773 ms, SE = 22 ms); the well-known priming of pop-out effect. There was
no significant Probe Response by Trial Type interaction (F < 1).
Mean error percentages were also submitted to an ANOVA that treated Probe Response and Trial Type as within-subject
factors. There was no main effect of Probe Response, F(1, 57) = 2.32, p = .133, g2p = .04, with errors being similar prior to ‘on-
task’ (M = 6.9, SE = 1.4) and ‘off-task’ (M = 8.3, SE = 1.1) reports. There was also no main effect of Trial Type, F(1, 57) = 2.49,
p = .120, g2p = .04, with errors being similar on target color repeat trials (M = 7.3, SE = 1.2) and switch trials (M = 8.0, SE = 1.3).
There was also no interaction between Probe Response and Trial Type (F < 1). Overall, the pattern of errors on the trials preced-
ing the thought probes is consistent with the overall pattern of RTs; there is no evidence that performance was in any way dif-
ferent prior to ‘off-task’ and ‘on-task’ reports.
In this final analysis, we adopt an individual differences approach in which we examine whether performance measures
(RT and accuracy) are predicted by overall rates of reported mind wandering (as per Hu, He, & Xu, 2012; Maillet & Rajah,
2013; McVay & Kane, 2009; Mrazek et al., 2012; Seli et al., in press; Thomson et al., 2013; Unsworth & McMillan, 2013;
etc.). That is, although we failed to find performance differences between on and off-task reports within individuals, it
remains possible that differences exist between individuals. To that end, we computed Pearson-product moment correlation
coefficients among proportion of off-task reports, RTs, and errors. There was no significant correlation between off-task
reports and either mean RT, df = 59, r = .016, p = .903, or mean error percentages, df = 59, r = .15, p = .257. Bivariate
correlations among off-task reports and RT/accuracy as a function of trial type (target color repeat/switch) also yielded no
significant relations.

Fig. 2. Individual correlations were calculated for each participant relating mind wandering to accuracy on repeat trials, accuracy on switch trials, response
time on repeat trials, and response time on switch trials. The mean Pearson correlation coefficients are shown (light grey bars). Error bars represent one
standard error of the mean.
 D.R. Thomson et al. / Consciousness and Cognition 27 (2014) 14–26 19

Fig. 3. Response times (A) and error percentages (B) on the five-trial runs preceding thought probes in Experiment 1. Data are shown as a function of
whether the target color on a given trial was the same (light grey) or different (dark grey) from the target color on the previous trial, and as a function of
whether the participant made an ‘on-task’ or ‘off-task’ report on the subsequent thought probe. Error bars represent one standard error of the mean.

2.3. Discussion

The present study sought to assess the impact (if any) of mind wandering on performance in a simple two-alternative
forced-choice singleton search task, both overall and over time. By examining the relation between mind wandering and per-
formance over time, we show that slopes representing changes in mind wandering over time significantly predict changes in
accuracy, but not RT over time. In other words, individuals with greater increases in mind wandering over time demonstrate
larger declines in accuracy over time. In addition, when examining average participant correlation coefficients (between
mind wandering and RT/accuracy), it becomes clear that the amount of mind wandering one engages in a given trial block
has a significant association with both RT and accuracy, indicating that within the individual, fluctuations in mind wandering
predict fluctuations in performance. Given that slopes representing changes in RT over block were not predicted by slopes
representing changes in mind wandering rates over blocks, yet RT was predicted by mind wandering from block to block, the
relation between mind wandering and RT seen here is not strictly linear. That is, linear increases or decreases in one measure
do not co-occur with linear increases or decreases in the other over time, yet fluctuations in one measure are mirrored by
fluctuations in the other (i.e. as one measure goes up or down from one block to the next, so does the other). Thus, there
is a strong relation between mind wandering and performance over time in the present task, which is especially noteworthy
given that singleton search requires little in the way of controlled processing (that is, the demands on attentional resources
are considered to be low) and therefore mind wandering might be expected to have no observable effects on performance (as
per Smallwood & Schooler, 2006).
Using a five-trial window prior to thought probes and binning RT and accuracy as a function of probe response, no effect
of probe response was observed for either measure. Furthermore, using an individual difference approach, no relation was
observed between overall reports of off-task thought and overall RT or accuracy. Therefore, using only established analytic
techniques for examining the effects of mind wandering on primary task performance, we would be forced to conclude that
in simple visual search, mind wandering has no observable impact. Our novel temporal-analytic approach however, reveals
that this is not the case. Before discussing the implications of these findings, we first seek to examine the temporal relation
between mind wandering and performance in a task in which the demands on attentional resources might be argued to be
considerably higher.

3. Study 2

In Study 1, temporal analyses of the effects of mind wandering on performance over time revealed the consequences of
off-task thought on behavior in both RT and accuracy in a simple visual search task. We now examine the temporal relation
between mind wandering and performance in yet another well-studied laboratory task, an Eriksen Flanker task (Eriksen &
Eriksen, 1974). In the Flanker task, participants must respond to the identity of a central item (in this case an arrow) while
ignoring flanking items that can either be congruent or incongruent with the central target. As a consequence, controlled
20 D.R. Thomson et al. / Consciousness and Cognition 27 (2014) 14–26

processing is needed to determine whether one can simply respond to the stimulus as a whole (as on congruent trials when
all items indicate the same response) or whether one must selectively attend to only the central item (as on incongruent
trials when flanking items indicate an incorrect response). As a result, compared to the singleton search task employed in
Study 1, the demands on attentional resources are considerably higher here. Whereas attentional guidance in the search task
employed in Study 1 can be achieved in an (arguably) entirely ‘bottom-up’ manner (see Theeuwes, 1991, 1992), the Flanker
interference task employed here requires the exertion of top-down control over responding (since the majority of ‘bottom-
up’ information will lead to the execution of an incorrect response on half of the trials). Therefore, as a result of the fact that
there will be relatively few ‘free’ attentional resources available for mind wandering in the Flanker task, mind wandering is
expected to have significant consequences on overall performance, even when assessed using standard analytic techniques
(since there is a greater probability that mind wandering will consume resources needed for the task, as per Smallwood &
Schooler, 2006, unlike in the search task employed in Study 1). We therefore predict that mind wandering during this task
will negatively impact overall performance, but will also be indicative of performance changes over time as in Study 1.

3.1. Method

3.1.1. Participants
67 undergraduates at the University of Waterloo (24 male, 43 female, mean age = 19.4 years) provided informed consent
and participated in exchange for course credit.

3.1.2. Stimuli and procedure

The experimental program was written in Python (www.python.org) and presented using PsychoPy software (Peirce,
2007). The experiment was conducted using the same set-up as in Study 1.
Participants were told that on each trial they would see five arrows arranged in a row in the center of the computer
screen. Importantly, on some trials the central arrow would point in the opposite direction from the four ‘flanking’ arrows,
whereas on the remaining trials the arrows would all point in the same direction. The task of the participant was to indicate
whether the central arrow pointed left (by pressing the ‘z’ key) or right (by pressing the ‘m’ key). Speed and accuracy were
stressed. As in Study 1, participants were informed that their thoughts would be probed at various times throughout the task.
Definitions and instructions for these probes were identical to those of Study 1.
Participants completed 1200 Flanker trials, of which half had ‘right-facing’ targets and half had ‘left-facing’ targets. In
addition, half of the trials were congruent (e.g. ‘>>>>>’) and half were incongruent (e.g. ‘<<><<’). These trial types were ran-
domly intermixed. Each trial began with the presentation of a central fixation cross that subtended 0.5° of visual angle and
was visible for 500 ms, after which the Flanker stimuli were presented. The central target appeared at fixation, and the whole
display subtended 5° of visual angle in horizontally, and 1° vertically. The Flanker display remained on the screen until a
response was made. Throughout the task, participants were required to respond to 40 thought probes. These probes were
presented pseudo-randomly, such that probes were separated by a minimum of 12 intervening search trials, and no probe
occurred during the first 15 trials of the experiment. The session lasted approximately 40 min.
In terms of analyses we assess the contribution of mind wandering (off-task responses to thought probes) to both
response time and accuracy. Response time (RT) is analyzed only for those trials in which a correct response was made.
In addition, we analyze data for RT and accuracy separately as a function of whether the target on a given trial was congruent
or incongruent with respect to the flanking arrows. We do this because it is well-known that RTs are significantly faster on
congruent trials relative to incongruent trials; the flanker interference effect (Eriksen & Eriksen, 1974). As in Study 1, we first
assess the relation between mind wandering and performance over time-on-task to address our primary theoretical goals,
before turning our attention to more standard, non-temporal analyses.

3.2. Results

3.2.1. Temporal analyses

In the following analyses we examine subjective reports of mind wandering as a function of trial-block. We define a trial
block as one-quarter of the total number of experimental trials (300). In order to examine the potential effects of mind wan-
dering on performance, we examine how mind wandering and performance measures (RT and accuracy) vary together over
time. We first examine mind wandering, RT and accuracy separately as a function of Block (see Fig. 4). An ANOVA that trea-
ted Block (1, 2, 3, 4) as a within subject factor revealed a significant effect of Block on the proportion of thought probes that
received an ‘off-task’ report, F(3, 198) = 29.99, p < .001, g2p = .31. Moreover, there was a significant linear trend, F(1, 66) = 78.61,
p < .001, g2p = .54, indicating that off-task reports increased in frequency as a function of time-on-task. Again, as in Study 1, a
closer inspection of Fig. 4a reveals that off-task reports level off between blocks three and four, indicating that there is likely
a ceiling on the proportion of time one will engage in off-task thought within the present task (once again, about 60% of the
time).
We conducted a similar ANOVA on RTs, but with both Block and Trial Type (congruent/incongruent) as within subject
factors. There was a significant main effect of Block on RT, F(3, 198) = 6.82, p < .001, g2p = .09, and a significant linear trend,
F(1, 66) = 4.31, p = .042, g2p = .06, indicating that RT decreased as a function of time-on-task. There was also a significant main
 D.R. Thomson et al. / Consciousness and Cognition 27 (2014) 14–26 21

Fig. 4. Proportion of ‘off-task’ reports (A), mean response times as a function of congruency (B), and proportion correct as a function of congruency (C) in
Experiment 2 are depicted as a function of time-on-task (block). Error bars represent one standard error of the mean.

effect of Trial Type, F(1, 66) = 189.70, p < .001, g2p = 74; the Flanker effect. Finally, there was no significant Block by Trial Type
interaction (F = 1.95).
The analysis of accuracy was identical to that of RT. There was a main effect of Block on accuracy, F(3, 198) = 34.28,
p < .001, g2p = .34, and a significant linear trend, F(1, 66) = 50.62, p < .001, g2p = .43, with accuracy decreasing as a function of
time-on-task. There was also a main effect of Trial Type, F(1, 66) = 143.15, p < .001, g2p = .68; the Flanker effect. There was also

Fig. 5. Individual correlations were calculated for each participant relating mind wandering to accuracy on congruent trials, accuracy on incongruent trials,
response time on congruent trials, and response time on incongruent trials. The mean Pearson correlation coefficients are shown (light grey bars). Error bars
represent one standard error of the mean.
22 D.R. Thomson et al. / Consciousness and Cognition 27 (2014) 14–26

a significant Block by Trial Type interaction, F(3, 198) = 9.62, p < .001, g2p = .13, and a significant linear trend, F(1, 66) = 30.25,
p < .001, g2p = .31, indicating that the Flanker effect increased in size as a function of time-on-task.
As in Study 1, in order to assess whether individuals with greater increases in mind wandering across block demonstrated
greater changes in performance across block, we next calculated slopes representing changes in the proportion of off-task
reports as a function of Block. We then compared these slopes to slopes representing changes in RT and accuracy over exper-
imental blocks. Pearson-product moment correlations reveal that changes in mind wandering over time-on-task marginally
(though not significantly) predict changes in accuracy, df = 66, r = .22, p = .074, and RT, n = 66, r = .21, p = .081.
Finally, as in Study 1, we compute individual Pearson correlation coefficients for each participant. That is, we correlate
mind-wandering rates across block with RT and accuracy across block for each participant. We then compute the mean r-
value for each series of correlations and compare this value to zero (which would indicate that overall, there is no relation).
We computed mean correlation coefficients between mind wandering and: congruent/incongruent RT, as well as congruent/
incongruent accuracy. One-sample t-tests revealed these means to be statistically different from zero for both congruent and
incongruent trial accuracy, t(65) = 6.56, p < .001, d = .81, t(66) = 6.43, p < .001, d = .79, but not RT (t’s < 1). These can be seen in
Fig. 5.

3.2.2. Non-temporal analyses

We first analyzed Mean RT and mean accuracy as a function of Probe Response and Trial Type using only the trials prior to
thought probes as in Study 1. These data are depicted in Fig. 6.
RTs were submitted to a repeated-measures ANOVA that treated Probe Response (on-task, off-task) and Trial Type (incon-
gruent, congruent) as within subject factors. There was no main effect of Probe Response (F = 1.12), with RTs being similar for
the trials leading up to ‘on-task’ (M = 530 ms, SE = 10 ms) and ‘off-task’ (M = 550 ms, SE = 15 ms) reports. There was a signif-
icant main effect of Trial Type, F(1, 66) = 50.73, p < .001, g2p = .44, with RTs being faster on congruent trials (M = 499 ms,
SE = 9 ms) than on incongruent trials (M = 580 ms, SE = 15 ms); the well-known Flanker effect. There was no significant Probe
Response by Trial Type interaction (F < 1).
Mean error percentages were also submitted to an ANOVA that treated Probe Response and Trial Type as within subject
factors. There was a significant main effect of Probe Response, F(1, 66) = 43.6, p < .001, g2p = .40, with fewer errors being com-
mitted prior to ‘on-task’ (M = 5.6%, SE = 0.5%) than ‘off-task’ (M = 11.2%, SE = 1.0%) reports. There was also a main effect of Trial
Type, F(1, 66) = 42.10, p < .001, g2p = .39, with fewer errors being committed on congruent trials (M = 5.4%, SE = 0.6%) than incon-
gruent trials (M = 11.4%, SE = 0.8%). There was no interaction between Probe Response and Trial Type (F < 1).
As in Study 1 we conducted an analysis at the level of individual differences, computing Pearson-product moment cor-
relation coefficients among proportion of off-task reports, RTs, and errors. There was no significant correlation between
off-task reports and mean RT, df = 66, r = .10, p = .433, however, the number of off-task reports provided by each participant
significantly predicted mean error percentages, df = 66, r = .25, p = .045.

Fig. 6. Response times (A) and error percentages (B) on the five-trial runs preceding thought probes in Experiment 2. Data are shown as a function of
whether the trial is congruent (light grey) or incongruent (dark grey), and as a function of whether the participant made an ‘on-task’ or ‘off-task’ report on
the subsequent thought probe. Error bars represent one standard error of the mean.
 D.R. Thomson et al. / Consciousness and Cognition 27 (2014) 14–26 23

3.3. Discussion

The purpose of Study 2 was to assess the link between mind wandering and performance over time in a task in which the
demands on attentional resources are considerably higher than in Study 1. The results were clear: mind wandering had a
strong impact on accuracy overall, and furthermore, fluctuations in mind wandering strongly predicted fluctuations in accu-
racy within the individual. That is, as in Study 1, the amount of mind wandering a participant reported within a given block
was predictive of performance within that block (again, this analysis is sensitive to non-linear changes in mind wandering
and performance over time, whereas the correlational analysis of slopes between individuals is not). Unlike in Study 1 how-
ever, here we observed a strong effect of mind wandering ‘in the moment’ using the probe trial-binning technique. As stated
previously, the Flanker task employed in the present study involves controlled (i.e. ‘top-down’) processing in a way that the
search task in Study 1 does not, and is therefore considerably more demanding of attentional resources. As a consequence,
particular instances of off-task thought are more likely to infringe on resources needed for the primary task (as per
Smallwood & Schooler, 2006), and thus, performance suffers. Crucially however, in both tasks used in the present work,
the overall amount of mind wandering reported within a given block was strongly predictive of performance, suggesting that
even if mind wandering does not have a measurable influence on performance ‘in the moment’ (as in Study 1), it (at the very
least) indexes more global changes in attention that are related to changes in performance.

4. General discussion

The primary purpose of the present work was to directly assess the relation between self-reported instances of off-task
thought (i.e. mind wandering) and performance over time in well-studied performance tasks. In a low-demand task (pop-out
search), no overall relation between mind wandering and performance was observed using standard analytic techniques,
whereas in a high-demand task, it was shown that mind wandering is associated with less accuracy. Importantly, temporal
analyses revealed that, in both tasks, changes in mind wandering over time strongly predicted changes in performance over
time, specifically at the individual subject level. Whereas prior work has only provided incidental evidence that changes in
mind wandering over time may be related to performance changes, here we provide direct evidence that fluctuations in off-
task thought are tightly linked to fluctuations in performance.

4.1. Theoretical implications

In their attentional-resource account of the wandering mind, Smallwood & Schooler, 2006 (see also Smallwood, 2010)
argue that on-task and off-task thoughts compete for the same limited pool of executive/attentional resources. It is argued
that if the resource demands of the primary task are sufficiently low, then mind wandering can unfold without impinging on
the resources needed for the primary task. Indeed, if one considers the analyses of performance prior to on- and off-task
reports, or the individual differences approach in Study 1 (the ‘standard’ analytic techniques), it would seem that this
supposition is largely supported. Mainly, performance did not differ prior to on- and off-task reports, nor did individual
differences in reports of overall mind wandering predict overall performance. However, when one considers the temporal
analyses conducted on the data from Study 1, it becomes clear that fluctuations in mind wandering from block to block
strongly predict fluctuations in both accuracy and response time, despite the fact that singleton search is one of the least
demanding tasks in terms of controlled processing (that is, it unfolds in a ‘bottom-up’ manner – see Theeuwes, 1991,
1992). It should be noted that while mean correlations between mind wandering and accuracy over time were negative,
so were mean correlations between mind wandering and RT, indicating that mind wandering may not necessarily impair
performance in a singleton search task in which speed and accuracy are equally stressed, but rather, that it may simply alter
the way in which speed and accuracy are traded-off by the observer. Indeed, the results of Study 1 suggest that this is a
perfectly reasonable interpretation. Nonetheless, the fact that mind wandering is associated with performance at all is a sig-
nificant finding, given the low demands on controlled processing. At present, leading theoretical accounts of mind wandering
do not explicitly allow for this and therefore, the present work serves to motivate further specification of existing theory in
that regard.
The present findings also have strong implications for theories of performance decrements over time in laboratory tasks.
Specifically, it has long been known that performance tends to decline over time-on-task in sustained attention/vigilance
tasks (Mackworth, 1948, 1950). Currently, researchers disagree on whether these performance declines owe to increases
in the frequency of off-task thought, or to something else entirely (such as the depletion of information processing resources;
Caggiano & Parasuraman, 2004; Helton et al., 2005; Smit et al., 2004). To date, there is no direct examination of the relation
between mind wandering and performance over time in well-studied performance tasks. The results of the present work are
a significant step forward in that regard by demonstrating not only that mind wandering rates do in fact increase over time-
on-task (which has been shown previously – Cunningham et al., 2000; McVay & Kane, 2012), but also that fluctuations in
mind wandering from block to block strongly predict performance from block to block (which has never been shown pre-
viously). These results suggest that changes in performance over time are tightly linked to increases in the frequency of
off-task thought, at least in the tasks employed here. To be clear, the present data show that performance from block to block
within an individual, is tightly linked to the amount of mind wandering that individual reports from block to block. We are
24 D.R. Thomson et al. / Consciousness and Cognition 27 (2014) 14–26

not saying that the effects of mind wandering on performance change in some way over time, although this is an intriguing
possibility and may represent a fruitful avenue for future work.

4.2. Methodological implications

To date, researchers interested in the relation between mind wandering and various performance measures have tended
to examine ways in which overall performance relates to mind wandering, and have largely overlooked the temporal prop-
erties of mind wandering. The most prominent analytic techniques used by mind-wandering researchers include: (1) within-
subjects analyses at the local level, and (2) between-subjects individual-differences analyses. Given the results of the present
study, it is likely that these techniques may over look more subtle influences of mind wandering on performance in certain
task contexts.
Analyzing the effects of mind wandering on behavior at the local level involves identifying trials during the experiment
that can, on the basis of some type of subjective report, be classified as either on-task or off-task periods. After this classi-
fication has been made, the researcher can then compare performance measures associated with both on- and off-task peri-
ods to determine whether performance differs as a function of these periods. For example, Seli, Carriere, et al. (2013) and Seli,
Cheyne, et al. (2013) presented thought probes during the Metronome Response Task (MRT), which allowed them to distin-
guish between subjectively reported periods of on-task and off-task thought. They then compared response variance on the
five trials preceding on- and off-task reports, demonstrating that response time variability is greater during periods of mind
wandering (using the same analytic technique, Bastian and Sackur (2013) also observed a relation between mind wandering
and response variance). Other researchers have successfully used this technique to expose the consequences of mind wan-
dering on: (1) motor-tracking ability (e.g., Kam et al., 2012), (2) reading comprehension (e.g., Dixon & Li, 2013), (3) oculo-
motor behaviors (e.g., Reichle, Reineberg, & Schooler, 2010; Smilek, Carriere, & Cheyne, 2010; Uzzaman & Joordens, 2011),
(4) response inhibition (e.g., Smallwood et al., 2007), (5) performance in visual-attention tasks (e.g., Kam, Dao,
Stanciulescu, Tildesley, & Handy, 2013), fidgeting behaviors (e.g., Seli et al., in press), and (6) driving performance (e.g.,
He et al., 2011; Yanko & Spalek, in press). On the other hand, using these same analytic techniques, no differences in perfor-
mance prior to on- and off-task reports were observed in (1) single word reading (Thomson et al., 2013; Experiment 1), (2)
Stroop performance (where congruent and incongruent trials were blocked – Thomson et al., 2013; Experiment 2), and sin-
gleton search (Study 1 of the present article).
The fact that ‘in-the-moment’ mind wandering did not affect performance (assessed via the probe trial-binning tech-
nique) in Study 1, yet the amount of mind wandering within a given block was strongly related to performance is notewor-
thy and deserves further comment. By including all trials within a given block in our within-subject correlations, we likely
had more power to detect the effects of mind wandering on performance. By comparison, the probe trial-binning technique
uses only the five trials prior to each thought probe and so discards a large proportion of the experimental trials. As a result,
the unique methodological approach taken here (i.e. within-participant correlational analyses) seems to be more sensitive to
the effects of mind wandering on performance than (at least some) pre-existing techniques.
Another analytic technique that is frequently used in studies of mind wandering involves correlating people’s perfor-
mance on a given task with their propensity to mind wander either during the task (i.e., state-level mind wandering) or
in everyday life (i.e., trait-level mind wandering). For example, Schooler, Reichle, and Halpern (2004), sought to determine
whether the propensity to mind wander during a reading task predicted performance on a subsequent comprehension test.
As predicted, Schooler and colleagues observed a significant negative relation between mind-wandering during the reading
task and comprehension of the reading material, indicating that the more people mind wander while reading, the poorer
their comprehension of the to-be-read material (see also Unsworth & McMillan, 2013). Other studies have successfully
employed the correlational approach to expose the relation between mind wandering and: (1) working-memory capacity
(e.g., Mrazek et al., 2012), (2) performance on encoding (Thomson et al., in press) and retrieval (Maillet & Rajah, 2013) of
information in memory, (3) response variability in a sustained-attention task (e.g., McVay & Kane, 2009), (4) fidgeting during
a sustained-attention task (Seli et al., in press), (5) performance on the Attention Network Test (ANT; e.g., Hu et al., 2012),
and (6) the tendency to engage in media multitasking (e.g., Ralph, Thomson, Cheyne, & Smilek, in press).
In summary then, the vast majority of research aimed at exposing the deleterious effects of mind wandering on primary
task performance has focused on either, (1) on the differences in performance prior to on- and off-task reports or, (2) the
relation between overall performance and overall reports of mind wandering. These procedures were applied to the present
experiments with limited success (particularly in Study 1). However, by capitalizing on the fact that both mind wandering
and performance change drastically as a function of time-on-task, we have been able to provide novel analytic techniques
with which to assess the effects of mind wandering on performance. Indeed, this temporal approach may be more sensitive
to the ‘subtle’ effects of mind wandering on performance in so-called ‘low demand’ tasks. Specifically, only our temporal
approach exposed the effects of mind wandering on performance in Study 1. We therefore argue that when researchers
are attempting to assess the consequences of mind wandering on performance, in future they make use of procedures that
will allow for a temporal analytic approach. That is, experimental designs that employ repeated thought-sampling (either
pseudo-randomly throughout the task, or regularly at the end of each ‘block’) will be well-suited for this purpose. This is
a particularly salient point given that some researchers who are explicitly interested in how mind wandering affects perfor-
mance over time, obtain subjective reports of mind wandering at only a single point in time (Helton & Russell, 2012; Helton &
Warm, 2008). Furthermore, even when standard analytic techniques do expose the influence of mind wandering on behavior
 D.R. Thomson et al. / Consciousness and Cognition 27 (2014) 14–26 25

(as in Study 2), the temporal approach outlined here may serve as a valuable complement with which to further delineate
the nature of the impact of mind wandering on performance over time.

4.3. Conclusions

Although anecdotal evidence suggests that our ability to pay attention in a given setting is strongly linked to the amount
of time we are required to attend, little empirical evidence exists to confirm the relation between mind wandering and per-
formance over time. In fact, some researchers have argued that mind wandering and performance fluctuations over time may
be independent of one another (Helton & Head, in press), at least in some task contexts. Here, using novel analytic tech-
niques, we demonstrate that mind wandering and performance over time are tightly linked. Moreover, this is the case even
in tasks in which we may feel that we can ‘get away’ with mind wandering, due to their phenomenological ease. In summary,
the present work suggests that, all other things being equal, the longer we try and focus our attention on a given external
task, the more likely we are to engage in mind wandering, and in turn, the greater the changes in behavior will be.

Acknowledgments

This work was supported by NSERC Discovery Grants awarded to D.S. and D.B. We would also like to thank Angela Mun
and Kirill Zaitsev for their help with data collection. In addition we thank two reviewers who provided helpful comments and
suggestions on an earlier version of this manuscript.

References

Bastian, M., & Sackur, J. (2013). Mind wandering at the fingertips: Automatic parsing of subjective states based on response time variability. Frontiers in
Psychology, 4, 573.
Caggiano, D. M., & Parasuraman, R. (2004). The role of memory representation in the vigilance decrement. Psychonomic Bulletin & Review, 11, 932–937.
Cunningham, S., Scerbo, M. W., & Freeman, F. G. (2000). The electrocortical correlates of daydreaming during vigilance tasks. Journal of Mental Imagery, 24,
61–72.
Dixon, P., & Li, H. (2013). Mind wandering in text comprehension under dual-task conditions. Frontiers in Psychology, 4, 682.
Eriksen, B. A., & Eriksen, C. W. (1974). Effects of noise letters upon the identification of a target letter in a nonsearch task. Perception & Psychophysics, 16,
143–149.
Feng, S., D’Mello, S., & Graesser, A. C. (2013). Mind wandering while reading easy and difficult texts. Psychonomic Bulletin & Review, 20, 586–592.
Forster, S., & Lavie, N. (2009). Harnessing the wandering mind: The role of perceptual load. Cognition, 111, 345–355.
He, J., Becic, E., Lee, Y. C., & McCarley, J. S. (2011). Mind wandering behind the wheel: Performance and oculomotor correlates. Human Factors: The Journal of
the Human Factors and Ergonomics Society, 53, 13–21.
Helton, W. S., Hollander, T. D., Warm, J. S., Matthews, G., Dember, W. N., Wallaart, M., et al (2005). Signal regularity and the mindlessness model of vigilance.
British Journal of Psychology, 96, 249–261.
Helton, W. S., & Warm, J. S. (2008). Signal salience and the mindlessness theory of vigilance. Acta Psychologica, 129, 18–25.
Helton, W. S., & Russell, P. N. (2012). Brief mental breaks and content-free cues may not keep you focused. Experimental Brain Research, 219, 37–46.
Hu, N., He, S., & Xu, B. (2012). Different efficiencies of attentional orienting in different wandering minds. Consciousness & Cognition, 21, 139–148.
Kam, J. W., Dao, E., Blinn, P., Krigolson, O. E., Boyd, L. A., & Handy, T. C. (2012). Mind wandering and motor control: Off-task thinking disrupts the online
adjustment of behavior. Frontiers in Human Neuroscience. http://dx.doi.org/10.3389/fnhum.2012.00329.
Kam, J. W., Dao, E., Stanciulescu, M., Tildesley, H., & Handy, T. C. (2013). Mind wandering and the adaptive control of attentional resources. Journal of
Cognitive Neuroscience, 25, 952–960.
Mackworth, N. H. (1948). The breakdown of vigilance durning prolonged visual search. Quarterly Journal of Experimental Psychology, 1, 6–21.
Mackworth, N. H., (1950). Researches on the measurement of human performance (Medical Research Council Special Report Series No. 268). Her Majesty’s
Stationery Office, London, England.
Maillet, D., & Rajah, M. N. (2013). Age-related changes in frequency of mind-wandering and task-related interferences during memory encoding and their
impact on retrieval. Memory, 21, 818–831.
Maljkovic, V., & Nakayama, K. (1994). Priming of pop-out: I. Role of features. Memory & Cognition, 22, 657–672.
Manly, T., Robertson, I. H., Galloway, M., & Hawkins, K. (1999). The absent mind: Further investigations of sustained attention to response. Neuropsychologia,
37, 661–670.
McVay, J. C., & Kane, M. J. (2009). Conducting the train of thought: Working memory capacity, goal neglect, and mind wandering in an executive-control
task. Journal of Experimental Psychology: Learning, Memory, and Cognition, 35, 196–204.
McVay, J. C., & Kane, M. J. (2012). Drifting from slow to ‘‘d’oh!’’: Working memory capacity and mind wandering predict extreme reaction times and
executive control errors. Journal of Experimental Psychology: Learning, Memory, and Cognition, 38, 525–549.
Mrazek, M. D., Smallwood, J., Franklin, M. S., Chin, J. M., Baird, B., & Schooler, J. W. (2012). The role of mind-wandering in measurements of general aptitude.
Journal of Experimental Psychology: General, 141, 788–798.
Peirce, J. W. (2007). PsychoPy—Psychophysics software in Python. Journal of Neuroscience Methods, 162, 8–13.
Ralph, B. C., Thomson, D. R., Cheyne, J. A., & Smilek, D. (in press). Media multitasking and failures of attention in everyday life. Psychological Research. http://
dx.doi.org/10.1007/s00426-013-0523-7.
Reichle, E. D., Reineberg, A. E., & Schooler, J. W. (2010). Eye movements during mindless reading. Psychological Science, 21, 1300–1310.
Riby, L. M., Smallwood, J., & Gunn, V. P. (2008). Mind wandering and retrieval from episodic memory: A pilot event-related potential study. Psychological
Reports, 102, 805–818.
Risko, E. F., Anderson, N., Sarwal, A., Engelhardt, M., & Kingstone, A. (2012). Everyday attention: Variation in mind wandering and memory in a lecture.
Applied Cognitive Psychology, 26, 234–242.
Robertson, I. H., Manly, T., Andrade, J., Baddeley, B. T., & Yiend, J. (1997). Oops!’: Performance correlates of everyday attentional failures in traumatic brain
injured and normal subjects. Neuropsychologia, 35, 747–758.
Schooler, J. W. (2004). Zoning out while reading: evidence for dissociations between experience and metaconsciousness. In D. T. Levin (Ed.). Thinking and
Seeing: Visual Metacognition in Adults and Children (pp. 203–226). MIT Press.
Seli, P., Carriere, J. S., Levene, M., & Smilek, D. (2013). How few and far between? Examining the effects of probe rate on self-reported mind wandering.
Frontiers in Psychology, 4, 430.
Seli, P., Cheyne, J. A., & Smilek, D. (2013). Wandering minds and wavering rhythms: Linking mind wandering and behavioral variability. Journal of
Experimental Psychology: Human Perception and Performance, 39, 1–5.
26 D.R. Thomson et al. / Consciousness and Cognition 27 (2014) 14–26

Seli, P., Carriere, J. S. A., Thomson, D. R., Cheyne, J. A., Martens, K. A. E., & Smilek, D. (in press). Restless mind, restless body. Journal of Experimental Psychology:
Learning, Memory, and Cognition. http://dx.doi.org/10.1037/a0035260.
Smallwood, J., McSpadden, M., & Schooler, J. W. (2007). The lights are on but no one’s home: Meta awareness and the decoupling of attention when the mind
wanders. Psychonomic Bulletin & Review, 14(527), 533.
Smallwood, J., Davies, J. B., Heim, D., Finnigan, F., Sudberry, M., O’Connor, R., et al (2004). Subjective experience and the attentional lapse: Task engagement
and disengagement during sustained attention. Consciousness and Cognition, 13, 657–690.
Smallwood, J., & Schooler, J. W. (2006). The restless mind. Psychological Bulletin, 132, 946–958.
Smallwood, J. (2010). Why the global availability of mind wandering necessitates resource competition: Reply to McVay and Kane (2010). Psychological
Bulletin, 136, 202–207.
Smilek, D., Carriere, J. S., & Cheyne, J. A. (2010). Out of mind, out of sight eye blinking as indicator and embodiment of mind wandering. Psychological Science,
21, 786–789.
Smit, A. S., Eling, P. A., & Coenen, A. M. (2004). Mental effort causes vigilance decrease due to resource depletion. Acta Psychologica, 115, 35–42.
Teasdale, J. D., Dritschel, B. H., Taylor, M. J., Proctor, L., Lloyd, C. A., Nimmo-Smith, I., et al (1995). Stimulus-independent thought depends on central
executive resources. Memory & Cognition, 23, 551–559.
Theeuwes, J. (1991). Exogenous and endogenous control of attention: The effect of visual onsets and offsets. Perception & Psychophysics, 49, 83–90.
Theeuwes, J. (1992). Perceptual selectivity for color and form. Perception & Psychophysics, 51, 599–606.
Thomson, D. R., & Milliken, B. (2012). Perceptual distinctiveness produces long-lasting priming of pop-out. Psychonomic Bulletin & Review, 19, 170–176.
Thomson, D. R., & Milliken, B. (2013). Contextual distinctiveness produces long-lasting priming of pop-out. Journal of Experimental Psychology: Human
Perception and Performance, 39, 202–216.
Thomson, D. R., Besner, D., & Smilek, D. (2013). In pursuit of off-task thought: Mind wandering-performance trade-offs while reading aloud and color
naming. Frontiers in Psychology, 4, 360.
Thomson, D. R., Smilek, D., & Besner, D. (in press). On the asymmetric effects of mind wandering on levels of processing at encoding and retrieval.
Psychonomic Bulletin & Review. http://dx.doi.org/10.3758/s13423-013-0526-9.
Treisman, A. M., & Gelade, G. (1980). A feature-integration theory of attention. Cognitive Psychology, 12, 97–136.
Unsworth, N., & McMillan, B. D. (2013). Mind wandering and reading comprehension: Examining the roles of working memory capacity, interest,
motivation, and topic experience. Journal of Experimental Psychology: Learning, Memory, and Cognition, 39, 832–842.
Uzzaman, S., & Joordens, S. (2011). The eyes know what you are thinking: Eye movements as an objective measure of mind wandering. Consciousness &
Cognition, 20, 1882–1886.
Yanko, M. R., & Spalek, T. M. (in press). Driving with the wandering mind the effect that mind-wandering has on driving performance. Human Factors: The
Journal of the Human Factors and Ergonomics Society. http://dx.doi.org/10.1177/0018720813495280.