Lost in thoughts: Neural markers of low alertness during mind wandering

Abstract

During concentration tasks, spontaneous attention shifts occurs towards self-centered matters. Little is known about the brain oscillatory activity underlying these mental phenomena. We recorded 128-channels electroencephalographic activity from 12 subjects performing a breath-counting task. Subjects were instructed to press a button whenever, based on their introspective experience, they realized their attention had drifted away from the task. Theta (4-7 Hz) and delta (2-3.5 Hz) EEG activity increased during mind wandering whereas alpha (9-11 Hz) and beta (15-30 Hz) decreased. A passive auditory oddball protocol was presented to the subjects to test brain-evoked responses to perceptual stimuli during mind wandering. Mismatch negativity evoked at 100 ms after oddball stimuli onset decreased during mind wandering whereas the brain-evoked responses at 200 ms after stimuli onset increased. Spectral analyses and evoked related potential results suggest decreased alertness and sensory processing during mind wandering. To our knowledge, our experiment is one of the first neuro-imaging studies that relies purely on subjects' introspective judgment, and shows that such judgment may be used to contrast different brain activity patterns.

Full text

!
1!
Lost in Thoughts: Neural Markers of Low Alertness During Mind Wandering
Claire BRABOSZCZa,b and Arnaud DELORMEa, b, c
Authors affiliation
a. Centre de Recherche Cerveau et Cognition, UMR 5549, Paul Sabatier University,
Faculté de Médecine de Rangueil 31062 Toulouse, Cedex 9, France
b. CERCO, CNRS, Toulouse, France
c. Swartz Center for Computational Neuroscience , University of California San
Diego ,9500 Gilman Dr Dept 0559 ,La Jolla CA 92093-0559, USA
Corresponding Author:
Adress: Claire Braboszcz Centre de Recherche Cerveau et Cognition, UMR 5549,
Université Paul Sabatier, Faculté de Médecine de Rangueil 31062 Toulouse, Cedex
9, France
Tel. : 0033 663 945 517
Fax : 0033 562 172 809
Keywords: alpha; attention; EEG; introspection; theta; auditory oddball
Abbreviations:
MW: mind-wandering; BF: breath focus; Odd: oddball stimulus; Std: standard
stimulus; ERSP: event related spectrum perturbation

!
2!
ABSTRACT
During concentration tasks, spontaneous attention shifts occurs towards self-
centered matters. Little is known about the brain oscillatory activity underlying these
mental phenomena. We recorded 128-channels electroencephalographic activity
from 12 subjects performing a breath-counting task. Subjects were instructed to
press a button whenever, based on their introspective experience, they realized their
attention had drifted away from the task. Theta (4-7Hz) and delta (2-3.5Hz) EEG
activity increased during mind wandering whereas alpha (9-11Hz) and beta (15-
30Hz) decreased. An auditory oddball protocol was presented to the subjects to test
brain-evoked responses to perceptual stimuli during mind wandering. Mismatch
negativity evoked at 100 ms after oddball stimuli onset decreased during mind
wandering whereas the brain-evoked responses at 200 ms after stimuli onset
increased. Spectral analyses and evoked related potential results suggest decreased
alertness and sensory processing during mind wandering. To our knowledge, our
experiment is one of the first neuro-imaging studies that relies purely on subjects'
introspective judgment, and shows that such judgment may be used to contrast
different brain activity patterns.
1. INTRODUCTION
While reading books, most people have had the experience of finding their
attention drifts towards self-centered matters. After some time (ranging from seconds
to minutes), the readers realize they are mind wandering and bring their attention

!
3!
back to their reading. Mind wandering episodes thus correspond to the emergence of
task-unrelated thoughts and affects that are attracting the attention away from the
task at hand (Smallwood and Schooler 2006; Mason, Norton et al. 2007). Not
surprisingly, mind wandering episodes occur in our everyday life quite often - for
instance, as soon as we perform a task and start realizing we are thinking about
something else while doing it. One may think that avoiding these attention shifts is
only a matter of concentration and willingness to carry out a mental task. Yet, after
weeks, months, or years of training in tasks involving sustained concentration ± such
as focused meditation practice - subjects realize that these events seem to just
happen, despite purposefully trying to avoid them - see Braboszcz (2010) for a
review of mind wandering during meditation practice.
The experience of mind wandering thus highlights the existence of moment to
moment subjectively-attested changes of attentional focus from a task to non-task
related thoughts and we believe that these changes would most likely be associated
with different brain activity. Although it is a common phenomenon, and although its
implication for consciousness research and the study of attention processes is
critical, the brain dynamics associated with mind wandering have not yet been
studied directly.
Mind wandering has been associated with lower level of alertness and
vigilance (Oken, Salinsky et al. 2006), a mental state with limited external information
processing where attention is decoupled from the environment (Smallwood and
Schooler 2006). Supporting this hypothesis, human subjects exhibited decreased
performance in rare-target oddball detection tasks during mind wandering (Giambra
1995). In addition, the amplitude of the P300 event-related potential component was
reduced during mind wandering, suggesting a decrease in attentional resources

!
4!
directed towards stimulus processing (Smallwood, Beach et al. 2008).
Although the brain dynamics associated with mind wandering have not been
studied, a number of studies have investigated the brain dynamics associated with
the resting state - an awake neutral state that is not associated with any specific
cognitive task and that is prone to mind wandering (Gusnard and Raichle 2001;
Mazoyer, Zago et al. 2001). Studies coupling both EEG and fMRI found that the
activity in different EEG frequency bands is spontaneously fluctuating at rest and can
be correlated to spontaneous fluctuations of the BOLD signal (Laufs, Holt et al. 2006;
Mantini, Perrucci et al. 2007). These fluctuations seem to underlie two distinct modes
of cerebral activity: a mode dominated by fast frequency waves (12-30Hz, beta) that
may index higher degrees of task-related attention (Ray and Cole 1985; Laufs, Holt
et al. 2006), and a mode dominated by slow 3-7Hz theta waves oscillations that has
been linked to decreased sustained task-related attention and diverse stages of
transition from wake to sleep (Loomis, Harvey et al. 1937; Makeig and Inlow 1993;
Klimesch 1999; Smit, Droogleever Fortuyn et al. 2005). Based on these results, we
hypothesized that task-unrelated attentional drifts ± i.e. mind wandering - would be
associated with decreased vigilance and increased delta and theta power.
It has also been shown that brain evoked response to external stimuli change
with the degree of vigilance or sleep stage. For example, the negative brain response
to the sensory detection of a sudden change in the flux of auditory perception called
mismatch negativity (Naatanen, Paavilainen et al. 2007) is reduced during the early
sleep stages and drowsiness (Lang, Eerola et al. 1995; Winter, Kok et al. 1995).
Since the mind wandering state should be associated with decreased vigilance, we
expect to observe a decrease in the mismatch negativity amplitude in the mind-
wandering state compared to the breath focus state.

!
5!
We designed an experiment allowing subjects to experience mind wandering
in conditions we believe to be as close as possible to the way they are experiencing it
in their daily life. We chose a simple concentration task - a silent breath counting task
- that only requires weak cognitive involvement from the subject, a characteristic
known to favor the induction of mind wandering (Giambra 1995; Cheyne, Carriere et
al. 2006). Simultaneously we presented frequent and rare pure-frequency auditory
stimuli that subjects were instructed to ignore, and we used these stimuli to assess
the evoked electrophysiological activity during the mind wandering and breath
concentration states.
2. MATERIALS AND METHODS
3DUWLFLSDQWV
Sixteen volunteers from the laboratory staff and local universities (8 females
and 8 males; age 19-36 years old, mean: 27 and standard deviation 5) gave written
consent to participate to the experiment. Participants stated that they were not taking
any substances or medications that could potentially affect their concentration nor
having histories of major psychological disorders or any auditory deficiencies. Before
starting the experiment, all participants read the instructions and had the possibility to
ask questions about the experiment before giving written consent to participate in the
experiment. As detailed below, 4 of the 16 participants had to be excluded because
they did not report enough mind wandering episodes.
3URFHGXUH
Participants sat in a dark room. We asked them to keep their eyes closed
during the recording session. Participants were instructed to count each of their
breath cycles (inhale/exhale) from 1 to 10. As subjects often lack immediate

!
6!
awareness of their mind wandering episodes (MWE), we could not ask them to signal
MWE occurrence at the moment their attention was drifting away from the task.
Instead, we asked them to indicate whenever they realized their attention had drifted,
that is whenever they gained meta-consciousness (Schooler 2002) of their mind
wandering episodes. We asked subjects to hold a button in their right hand and press
it whenever they became aware of having lost track of their breath count. The
following instructions were given to subjects to define what was meant by losing track
RIWKH FRXQW³\RX VWRSSHGFRXQWLQJ´ ³\RXFRXQWHG RYHU EUHDWKV´RU³\RX KDGWR
reflect intensively to fiJXUH RXW ZKDW ZDV WKH QH[W FRXQW´ 2QFH WKH\ SUHVVHG WKH
button, participants were instructed to bring their focus back to their breath and start
counting again from one. We read task directions to participants and made sure they
understood them.
The experiment lasted about one hour and 10 minutes, split into three blocks
of 20 minutes separated by five minutes of rest. At the end of each block, we asked
VXEMHFWV WR HYDOXDWH WKHLU DOHUWQHVV OHYHO GXULQJ WKH SDVW VHVVLRQ ³'LG \RX IHHO OLNH
IDOOLQJ DVOHHS"´ WKHLU H\H PRYHPHQW DFWLYLW\³'LG \RX RSHQ \RXU H\HV ± if yes can
\RXHVWLPDWHKRZPDQ\WLPHV´DQGWKHLUVXEMHFWLYHH[SHULHQFHZKHQUHSRUWLQJPLQG-
ZDQGHULQJ HSLVRGHV ³'LG \RX SUHVV WKH EXWWRQ" ,I \HV ZKDW ZDV \RXU VXEMHFWLYH
H[SHULHQFH"´ None of the participants reported systematically opening their eyes
and none of the participants reported falling asleep. However, 6 of the 12 selected
participants reported some level of drowsiness at one time or another during the one-
hour experiment (see Discussion).
$XGLWRU\VWLPXOL
While performing the breath counting task, subjects were also presented with

!
7!
a passive auditory oddball protocol that they were instructed to ignore. The auditory
oddball protocol was composed of pure sounds of 500 Hz for the standard stimuli
(80% of the stimuli) and 1000Hz for the oddball (20% of the stimuli). Each sound
lasted 100 ms with 10 ms linear amplitude rising and falling times. Inter-stimulus
intervals randomly varied between 750 and 1250 ms. Oddball stimuli presentation
was pseudo-random to ensure there were never two oddball stimuli presented
successively. Auditory stimuli were calibrated at 72dB and played through a
loudspeaker located at 1.20 meters in front and 45 degrees on the right of the
subject.
5HFRUGLQJ
We recorded data using a 128-FKDQQHO :DYHJXDUG FDS $GYDQFHG 1HXUR
Technology Company ± ANT) out of which we used 124-channels - electrodes AFZ,
PO6, TP7, DQG32ZHUH GDPDJHG DQG OHIWRXW:HSOXJJHGWKH:DYHJXDUGFDS
into two synchronized 64-channel EEG amplifiers also from the ANT Company. We
kept most electrode impedances below 5KOhm although about 10% of the
electrodes still had higher impedance at the end of preparation - all impedances were
kept below 20Kohm as recommended in ANT ASA 4.0 software uVHU¶VJXLGH± ANT
recommendation is higher than the standard 5 Kohm because of the high impedance
of its amplifier. We used M1 mastoid electrode as reference and sampled the data at
1024Hz. We also recorded EKG by placing two bipolar electrodes on each side of the
VXEMHFW¶VWRUVR
$UWLIDFWVFRUUHFWLRQ
We first removed bad electrodes ± from 2 to 17 bad electrodes per subject.

!
8!
We then manually pruned the continuous data from non-stereotyped, unique artifacts
such as paroxysmal muscles activity - high frequency activities with large amplitude
over all electrodes - as well as electrical artifacts resulting from poor electrode
contacts - short-lasting aberrant oscillatory activity localized at a few electrode sites.
We then used Infomax Independent Component Analysis (Infomax ICA) on the
pruned data to reject artifacts. For each subject, we visually identified and rejected
one to five well-characterized ICA components for eye blink, lateral eye movements,
and temporal muscle noise (Delorme, Sejnowski et al. 2007). We used visual
inspection of component scalp maps, power spectrum and raw activity to select and
reject these artifactual ICA components.
'DWDSURFHVVLQJ
Data processing was performed under Matlab 7.0 (The Mathwork, Inc.) using
the EEGLAB 7.x toolbox (Delorme and Makeig 2004). We first downsampled the
EEG data from 1024 Hz to 256Hz and performed high-pass filtering at 1Hz using a
non-linear elliptic filter. In addition, we applied an elliptic non-linear notch filter
between 45 and 55 Hertz. For each subject, we then segmented the EEG data into
20-second GDWD HSRFKV FHQWHUHG RQ VXEMHFWV¶ EXWWRQ SUHVVHV :H FRQVLGHUHG WKDW
participants were mind wandering during the 10-second period that preceded the
button press and we considered that participants were concentrating on their breath
during the 10-second period that followed the button press (Christoff, Gordon et al.
2009). Four subjects did not have enough clean data epochs to be considered for
further analysis - the four subjects had six, five, five and one clean epochs
respectively. All the selected subjects had between 13 and 52 of such 20-second
clean EEG data epochs (mean of 30 per subject; standard deviation of 14), ensuring

!
9!
that, for each subject, there would be at least 20-30 stimuli in each condition to
compute ERPs (Kappenman and Luck 2010) ± see ERP analysis below. The total
number of analyzed mind wandering event across all subjects was 358.
For each of the two conditions, mind wandering and breath focus, we also
extracted data epochs from one second before to two seconds after the presentation
of auditory stimuli. So that auditory stimuli do not occur too close to a button press,
we removed all three-seconds data epochs containing a button press - thus button
presses were at least one second prior to the stimulus or at least two seconds after
the stimulus. This procedure ensured that the brain activity related to the button
press does not contaminate our analysis. In addition, we processed brain activity
from electrodes (Oz, Fz) that were not over pre-motor and motor regions limiting
potential contamination of button press brain related activity. We thus obtained four
groups of data epochs ± oddball and standard stimuli defined over two conditions:
mind wandering and breath focus. We computed mean event related potential (ERP)
using a -300 to 0 ms baseline and we performed ERP visualization after applying a
30Hz linear low pass filter - note that we used the non-filtered data for computing
statistics. We counted a total of 4326 standard stimuli (mean of 180 per subject;
standard deviation of 101) and 1040 oddball stimuli (mean of 43 per subject;
standard deviation of 23).
((*WLPHIUHTXHQF\DQDO\VLV
We applied Morlet wavelet decomposition (Goupillaud, Grossman et al. 1984)
to both the 20-second long data epochs time-locked to button presses and the short
3-second data epoch time-locked to auditory stimuli. We used 200 linearly-spaced

!
10!
time points and a series of 100 log-spaced frequencies ranging from 1 Hz to 100 Hz,
with 1.5 cycle at the lowest frequency increasing linearly and capping at eight cycles
at 30 Hz. For long 20-second epochs, we visualized absolute log power -
10*log10(X), X being absolute power at a given time-frequency point. For short three-
second epochs time-locked to auditory stimuli presentation, we also removed
baseline spectral activity by subtracting the pre-stimulus average baseline log-power
at each frequency (Delorme and Makeig 2004) .
6WDWLVWLFV
Statistical tests were performed on ERPs, time-frequency maps and
topographic maps using two-tailed paired parametric student t-test (df=11). Since
most representation involves hundreds of tests, correction for multiple comparisons
was performed using the Montecarlo and the cluster method as developed by Maris
(2007). This method first measures the extent of 1-D (length) or 2-D (surface) of
significance regions (uncorrected) and then tests if the extent of these regions is
significant using a Monte-Carlo approach. For channel topographies, we set the
number of channel neighbors to 4.5 before running Maris (2007) Matlab function. We
also tried FDR (False Discovery Rate) (Benjamini and Yekutieli 2001) to correct for
multiple comparisons and obtained similar results compared to the cluster method.
3. RESULTS
((*DFWLYLW\WLPHORFNHGWRPHWDFRQVFLRXVQHVVHYHQWV
The time frequency analysis of EEG data time-locked to meta-consciousness
event ± button press - VKRZVDVLJQLILFDQWLQIOXHQFHRIWKHVXEMHFW¶VDWWHQWLRQDOVWDWH

!
11!
at all frequency bands from 2 to 25 Hz (Fig. 1). The most pronounced state-
associated change on the EEG spectral activity occurs in the theta (4-7Hz) band
where absolute spectral power is significantly higher in the mind wandering state
compared to the breath focus state. This effect is observed at all electrode sites and
is larger over occipital and parieto-central regions. Absolute power in the delta band
(2-3.5Hz) showed the same trend although the largest power difference was now
observed over the fronto-central region. By contrast occipital alpha (9-11 Hz) and
fronto-lateral beta (15-30 Hz) power was significantly lower in the mind wandering
state compared to the breath focus state.
During the transition associated with the meta-consciousness (MC) event, the
alpha frequency band in Fig. 1 is not only affected in terms of amplitude but also in
terms of peak frequency. The peak frequency appears to increase by about 1 Hz
after the meta-conscious event for a period of about 2 seconds. To test if this
observation was significant across subjects, we defined three time windows, W1 from
-6 to -4 second before the MC event; W2 from 0 to 2 seconds after the MC event; W3
from 6 to 8 seconds after the MC event. For each subject and for each time window,
we then manually assessed the alpha peak frequency by taking the frequency of
maximum power between 8 and 12 Hz on the power spectrum ± the power spectrum
was computed by averaging log-power values of Fig. 1 over the windows of interest
W1, W2 and W3. Note that the alpha peak frequency could not be found for one of
the 12 subjects so we computed statistics using 11 subjects only. Bootstrap statistics
revealed significant difference between the central W2 window and the flanking W1
and W3 windows (W1 versus W2, p<0.0005, df=10; W3 versus W2, p<0.002, df=10)
but not between W1 and W3. Supplementary Fig. 1 is a movie showing the
dynamical change in the power spectrum where the alpha amplitude changes and

!
12!
peak frequency shifts are made clearly visible.
6WLPXOXVHYRNHGDFWLYLW\GXULQJPLQGZDQGHULQJDQGEUHDWKIRFXV
We first tested if the attentional state affected grand average ERPs of the
auditory stimuli in the passive oddball paradigm. We observed that the ERP positive
component at about 200 ms after stimulus presentation (P2) is significantly higher
over fronto-central sites from 180 to 280 ms during mind wandering than during
breath focus for both standard and oddball stimuli (Fig. 2). We did not observe any
significant interaction between mental state and type of stimuli in this latency range.
However, we did observe such an interaction at earlier latencies.
We found a significant effect of the type of stimulus - oddball or standard - on
the amplitude of the early ERP negative component between 90 and 120 ms after
stimulus onset (Fig. 3). After presentation of an oddball stimulus the ERP is
significantly more negative over frontal and temporal regions than after presentation
of a standard stimulus both in the breath focus and mind wandering conditions (Fig.
3C and 3D). This increased negativity for oddball is usually termed mismatch
negativity (Naatanen, Paavilainen et al. 2007). The mismatch negativity (MMN) was
larger during breath focus compared to mind wandering over the right frontal region
(Fig. 3E). Supplementary Fig.2 shows single subject average ERP values and
standard error for both the 180 to 200 ms and 90 to 120 ms ERP range.
We then investigated event-related activity using time-frequency
decompositions. The event-related spectral perturbation plot reveals increased theta
band power (4-7Hz) and decreased high alpha (10-14 Hz) and high beta (20-25Hz)
band power after stimulus presentation (Fig. 4). In general, statistical inference
testing between the mind wandering and the breath focus state returned a lower p-

!
13!
value for standard stimuli compared to oddball stimuli ± it might be a matter of
number of observations since there was, on average, five times more trials for
standard than for oddball stimuli. From 100 to 300 ms after standard auditory stimuli
presentation, theta (4-7 Hz) power was significantly higher on frontal sites when
subjects were mind wandering compared to when they were focusing on their breath.
Delta (2-3.5 Hz) power 200 to 350 ms after standard auditory stimulation follows the
same trend and we also observed a significant power increase for oddball stimuli at
occipital and frontal sites. High beta (20-25Hz) power from 100 to 300 ms after
standard stimuli presentation is significantly higher on parieto-occipital sites during
mind wandering compared to during breath focus. Interestingly, despite large high
alpha (10-14 Hz) evoked power to both standard and oddball stimuli, we did not
observe any significant effect of the attentional state on the ERSP in this frequency
band.
We also tested for difference of ERSP between standard and oddball stimuli
during both the mind wandering and the breath focus states. Only beta band power
from 100 ms to 300 ms after stimulus presentation differed significantly, being lower
for oddball stimuli (not shown). This effect was not significantly different between the
mind wandering and the breath focus states.
4. DISCUSSION
Our study aimed at characterizing the neural correlates of spontaneous and
task-unrelated mental activity (i.e. mind-wandering) and its effect on sensory
processing. Compared to a breath-focus mental state, we have shown that mind
wandering is characterized by a power amplitude increase in the theta frequency
band and a power amplitude decrease in the alpha and beta frequency bands. We

!
14!
also showed that, during mind wandering, standard auditory stimuli induce a higher
power in the theta and delta frequency band over parieto-occipital regions and higher
power in the high beta frequency band over frontal regions. The study of mean
evoked related potentials revealed that the amplitude of the P2 positive ERP
component is larger during mind wandering than during breath focus and that the
MMN is of smaller amplitude during mind wandering than during breath focus. Taken
together these results establish a strong link between the subjects¶ LQWHUQDO
experience ± mind wandering or breath focus ± and distinct neural correlates.
&RQWUROWDVNDQGQRYHOW\RIWKHH[SHULPHQWDOGHVLJQ
The control task being used to study mind wandering was critical. We chose a
breath focus task, which is a relatively neutral non-cognitive task. Ideally, one would
study mind wandering during several control attention engaging tasks. Here, we want
to emphasize the difficulty and novelty of the experimental design and why it was
impractical for us to use multiple control tasks. Other studies of mind wandering often
use tasks where subjects have to respond continuously to stream of stimuli (e.g.
(Smallwood, Beach et al. 2008). By contrast, in our task, we asked subjects to press
a button based on pure introspection. We wanted to collect as many behavioral
responses as possible, but despite one hour of recording for each subject, we only
obtained 13-52 clean data epochs per subject. In addition, despite the instruction to
stay still, subjects tended to have muscle artifacts in their EEG after pressing the
button, forcing us to reject about 25% of the data. Finally, subjects varied widely in
their propensity to report mind-wandering events, and four subjects had to be
excluded because they provided too few responses. This experiment was the first of
its kind and is a proof of concept that, despite the difficulty encountered, this type of
study is possible.

!
15!
%UDLQVWDWHVIOXFWXDWLRQDWUHVWDQGVSRQWDQHRXVUHSRUWVRIPLQGZDQGHULQJ
The difference in EEG activity between mind wandering and breath focus is
consistent with the Laufs (2006) EEG-fMRI study showing that spontaneous EEG
DOWHUQDWHV EHWZHHQ WZR VWDWHV DW UHVW 2XU VWXG\ UHILQHV /DXIV¶ UHVXOWV E\ VKRZLQJ
that the low frequencies (theta-delta) and high frequencies (alpha-beta) changes he
observed may be associated to a transition between a state of concentration on
processing external stimuli and involuntary mind wandering. Our EEG study further
confirms that when subjects are engaged in a task, the brain can spontaneously shift
into another alertness mode, which is most likely mind wandering.
fMRI bold signal during the resting state shows spontaneous fluctuations
between a "task positive" network comprising brain areas activated during attention-
demanding tasks and a "task negative" (or "default") network being activated during
rest and deactivated during these tasks (Fox, Snyder et al. 2005). Preceding reports
of mind wandering, Christoff (2009) found increased BOLD activity both in the default
network (precuneus, ventral anterior cingulate cortex and temporoparietal junction)
and in the frontal executive network. This result is consistent with an fMRI study
showing increased amount of mind wandering linked to increased amount of activity
in the task negative network (Mason, Norton et al. 2007). Continuous increase of
BOLD activity in the occipital, frontal and temporal parts of the defaults network is
also found during the transition from eyes-closed wakefulness to sleep (Olbrich,
Mulert et al. 2009). The higher occipito-parietal theta and fronto-central delta during
mind wandering could thus be related to increased BOLD activity in these areas.
Additional combined EEG-fMRI studies would be needed to establish a clearer link
between EEG and BOLD signature of the mind wandering state and its relation to the
default mode network, in particular regarding localization of the neuronal sources of

!
16!
the EEG rhythm correlated with the BOLD signal, .
We also observed a delta power increase during mind wandering, an increase
that we believe could be linked to decreased alertness. Spontaneous delta power
increase has been linked to decreased performance during cognitive processing
(Harmony, Fernandez et al. 1996). Spontaneous delta power increase has also been
associated with decreased level of alertness in various experimental setups (Makeig
and Inlow 1993; De Gennaro, Ferrara et al. 2001; Caldwell, Prazinko et al. 2003).
Moreover, as reviewed by Laufs (2006) and shown in this report, delta power
increase is associated with alpha power decrease, which has been associated with
low stages of vigilance (Loomis, Harvey et al. 1937; Roth 1961). Note that 6 of the 12
subjects reported some level of drowsiness during the experiment. To be sure that
our result did not pertain to drowsiness, Supplementary Fig. 3 shows the same time-
frequency decomposition as Fig. 1 although it only includes subjects that did not
report any drowsiness. The time-frequency patterns of Supplementary Fig. 3 are
almost identical to the one visible in Fig. 1.
The meta-consciousness event allowing the transition from the mind
wandering to breath focus is finally marked by a transient increase of about 1Hz of
the alpha peak frequency and also by a more long lasting increase in alpha power.
Re-directing the attention to the task requires increase working memory activity and
that has been shown to be correlated with alpha power increase (Jensen, Gelfand et
al. 2002), a power increase that may index re-activation of thalamo-cortical pathways
(Schreckenberger, Lange-Asschenfeldt et al. 2004). The alpha-peak frequency
increase may also be a marker of the attentional switch between mind wandering and
the focused task since Angelakis (2004) suggests that increase of peak alpha
frequency might represent a state of "cognitive preparedness".

!
17!
001DWWHQWLRQDOHUWQHVVDQGPLQGZDQGHULQJ
The study of evoked related potential shows an increased negativity at frontal
electrode sites for the ERP of oddball compared to the ERP of standard stimuli from
90 to 120 ms after stimulus presentation. This result corresponds to the mismatch
negativity (MMN) usually described as negative brain response to the sensory
detection of a sudden change in the flux of auditory perception (Naatanen R et al.,
2007). The MMN typically occurs approximately 100 to 150 ms after stimulus
presentation and is centered on fronto-central electrodes sites (Naatanen,
Paavilainen et al. 2007). The amplitude of the MMN is modulated by the direction of
WKHVXEMHFW¶VDWWHQWLRQ(Sabri, Liebenthal et al. 2006), and is larger when the attention
of the subject is directed toward the auditory stimuli (Alain and Woods 1997). Our
results show the MMN amplitude is lower during mind wandering compared to breath
focus, which suggest a disengagement of the attention from stimuli processing during
mind wandering.
The reduction of the MMN is also characteristic of drowsiness and the early
sleep stages (Sallinen and Lyytinen 1997; Nittono, Momose et al. 2001), which
supports the idea that the mind wandering state is associated with decreased
vigilance. This also suggests that mind wandering may share common traits with the
decreased alertness characterizing the transition from wake to sleep (Sabri, Labelle
et al. 2003). Note that we did not observe in our data the late negativity at about 300
ms after stimulus presentation that accompanies advanced states of drowsiness
leading to sleep or sleep itself (Winter, Kok et al. 1995; Campbell and Colrain 2002).
This suggests that our subjects were not deeply drowsy. Mind wandering could thus
correspond to an early state of drowsiness of decreased alertness and vigilance.

!
18!
/DWHVWLPXOXV HYRNHG DFWLYLW\DQG GLVHQJDJHPHQWRIDWWHQWLRQIURPVWLPXOL
SURFHVVLQJGXULQJPLQGZDQGHULQJ
ERPs analysis also reveals that the amplitude of the positive component at
about 200 ms (P2) is larger during mind wandering than during breath focus. This
effect is also present, although to a lesser extent, in Cahn (2007) who found a P2
component larger for distracting stimuli when subjects were actively reactivating
autobiographical memories ± which may be considered similar to mind wandering -
compared to when they were practicing meditation. Increase of the P2 component to
DXGLWRU\ VWLPXOXV KDV DOVR EHHQ DVVRFLDWHG ZLWK WKH GLVHQJDJHPHQW RI VXEMHFWV¶
attention toward stimuli (Naatanen and Picton 1987) and is also characteristic of the
sleep onset period (Campbell and Colrain 2002). Again, this result is consistent with
attentional disengagement toward stimuli processing during mind wandering.
We did not observe a P300 ERP component associated with the presentation
of the rare stimuli in our passive auditory oddball task. P300 is best observed in
active experimental design where the subjects have to respond to rare targets and is
usually hardly visible in passive oddball paradigms (Cahn and Polich 2008). Using an
active task, Smallwood (2008) showed a reduction of the P300 ERP during mind
wandering. Consistent with our MMN and P2 results, Smallwood (2008) result
suggest a disengagement of attention towards external stimuli processing.
The study of ERSP is harder to interpret since it is rarely presented in
literature. Increased evoked theta frequency over frontal regions may be related to
increased autobiographical memory engagement during mind wandering (Jensen
and Tesche 2002; Onton, Delorme et al. 2005). Note that the ERP differential scalp
maps from 180 to 280 ms were similar to the theta frequency maps with strong

!
19!
changes over occipital regions. The ERP is a complex combination of stimulus-
locked phase synchronization and spectral amplitude increase (Makeig, Westerfield
et al. 2002; Delorme, Westerfield et al. 2007). We tested if ERP and ERSP activities
were linked by computing the correlation between the ERSP activity and the ERP at
electrode site Fz for the evoked delta, theta, alpha and beta frequency band activity
shown in figure 2. We did not find any correlation between the early ERP negative
component between 90 and 120 ms after stimulus onset and any of the ERSP
components. However, when pooling data for both types of stimulus and both
attentional states, we did find a positive correlation (p<0.001; df=47; paired t-test)
between the evoked delta (2.5 to 3.5Hz) and high alpha (10-14Hz) activity 100 to 300
ms after stimulus onset and the late ERP positive component at 180 to 280 ms. This
indicates that both the late ERP complex and delta ERSP activity may index similar
processes in our passive auditory oddball task.
)XQFWLRQRIPLQGZDQGHULQJ
The functional role of mind wandering remains debated in philosophy and
experimental psychology. The concept of mind wandering plays an important role in
Buddhist psychology (Trungpa 2004) since it is a major obstacle to concentrative
meditation practice. Buddhist psychology argues that mind wandering is a non-
productive ego-FHQWHUHG VWDWH D VWDWH RI ³VOHHS´ ZKHUH RXU XQFRQVFLRXV FRQVWDQWO\
rehash the same thoughts and beliefs creating confusion and strengthening our
sense of self. By contrast, some researchers have suggested that mind wandering
may be useful to provide creative insight (Christoff, Gordon et al. 2009) in a way
similar to sleep-induced insight (Wagner, Gais et al. 2004). Our result of finding mind
wandering to be a state of low alertness supports both views. It can be considered a
VWDWHRIORZFRQFHQWUDWLRQRUUHODWLYH³VOHHS´DVDUJXHGLQ%XGGKLVWSV\FKRORJ\EXWLW

!
20!
could also be seen as an hypnagogic state that may lead to creative insights
(Boynton 2001). We believe that by studying the common brain structures and
dynamics involved in mind wandering, meditation, self, and creativity, brain-imaging
techniques could help bring new light to this debate.
Based on our results and previous studies, we conclude that mind
wandering is a low-alertness state of rest. If mind wandering corresponds to a state
of rest, one hypothesis is that subjects who are sleep deprived might spend more
time mind wandering during the day. The time that a subject spends mind wandering
may be estimated using probe-caught mind wandering techniques. Smallwood
(2006) place a distinction between self-caught and probe-caught mind wandering
episodes. Self-caught mind wandering is the type of mind wandering studied in this
report. By contrast, to assess the amount of time subjects spend mind wandering
while being unaware of it, they may be probed at regular intervals about their state of
mind wandering. We would thus anticipate that probe-caught mind wandering
frequency would increase with the amount of sleep deprivation. Finally, if the activity
in the default network is linked to mind wandering as previously claimed (Mason,
Norton et al. 2007; Sonuga-Barke and Castellanos 2007), we would expect that the
activity in the default network during the day would also increases with sleep
deprivation. Further studies should be able to verify or disprove these hypotheses.
In conclusion, we have shown the neurophysiologic markers of mind
wandering. Based only on subjective reports about mind wandering, we have
established that two different attentional states correspond to two distinct brain states
underlying different modes of sensory processing. Our results suggest that mind
wandering correspond to a state of rest, a state of low vigilance where stimulus
evoked responses are reduced. This study is one of the first event-related

!
21!
neuroimaging study to rely only on behavioral responses based on pure ± not
stimulus induced - introspective subjective reports. It further demonstrates that
neuro-phenomenological approaches to the study of subjective experience are
possible in neuroscience (Lutz and Thompson 2003) yet argues for the need of a
more fine-grained taxonomy of private mental states.
Acknowledgements: This project was supported by a small grant from the Mind and
Life foundation and by a PhD fellowship from the French CNRS governmental
organization. We also wish to thank Dr. Emmanuel Barbeau for his suggestions on
the manuscript.
REFERENCES
Alain, C. and D. L. Woods (1997). "Attention modulates auditory pattern memory as
indexed by event-related brain potentials." Psychophysiology 34(5): 534-46.
Angelakis, E., J. F. Lubar, et al. (2004). "Peak alpha frequency: an
electroencephalographic measure of cognitive preparedness." Clin
Neurophysiol 115(4): 887-97.
Benjamini, Y. and D. Yekutieli (2001). "The control of the false discovery rate in
multiple testing under dependency." Annals of Statistics 29(4): 1165 - 1188.
Boynton, T. (2001). "Applied Research Using Alpha/Theta Training for Enhancing
Creativity and Well-Being " Journal of Neurotherapy 5(1&2): 5-18.
Braboszcz, C., S. Hahusseau, et al. (2010). Meditation and Neuroscience: from basic
research to clinical practice. Handbook of Integrative Clinical Psychology,
Psychiatry, and Behavioral Medicine: Perspectives, Practices, and Research.
R. A. Carlstedt, Springer Publishing Co Inc.
Cahn, B. R. (2007). Neurophysiologic Correlates to Sensory and Cognitive
Processing in Altered States of Consciousness Neuroscience. San Diego,
UCSD. Ph.D: 294.
Cahn, B. R. and J. Polich (2008). "Meditation (Vipassana) and the P3a event-related
brain potential." International Journal of Psychophysiology In press.
Caldwell, J. A., B. Prazinko, et al. (2003). "Body posture affects
electroencephalographic activity and psychomotor vigilance task performance
in sleep-deprived subjects." Clin Neurophysiol 114(1): 23-31.

!
22!
Campbell, K. B. and I. M. Colrain (2002). "Event-related potential measures of the
inhibition of information processing: II. The sleep onset period." Int J
Psychophysiol 46(3): 197-214.
Cheyne, J. A., J. S. Carriere, et al. (2006). "Absent-mindedness: Lapses of conscious
awareness and everyday cognitive failures." Conscious Cogn 15(3): 578-92.
Christoff, K., A. M. Gordon, et al. (2009). "Experience sampling during fMRI reveals
default network and executive system contributions to mind wandering." Proc
Natl Acad Sci U S A 106(21): 8719-24.
De Gennaro, L., M. Ferrara, et al. (2001). "The boundary between wakefulness and
sleep: quantitative electroencephalographic changes during the sleep onset
period." Neuroscience 107(1): 1-11.
Delorme, A. and S. Makeig (2004). "EEGLAB: an open source toolbox for analysis of
single-trial EEG dynamics including independent component analysis." J
Neurosci Methods 134(1): 9-21.
Delorme, A., T. Sejnowski, et al. (2007). "Enhanced detection of artifacts in EEG data
using higher-order statistics and independent component analysis."
Neuroimage 34(4): 1443-9.
Delorme, A., M. Westerfield, et al. (2007). "Medial prefrontal theta bursts precede
rapid motor responses during visual selective attention." J Neurosci 27(44):
11949-59.
Fox, M. D., A. Z. Snyder, et al. (2005). "The human brain is intrinsically organized
into dynamic, anticorrelated functional networks." Proc Natl Acad Sci U S A
102(27): 9673-8.
Giambra, L. M. (1995). "A laboratory method for investigating influences on switching
attention to task-unrelated imagery and thought." Conscious Cogn 4(1): 1-21.
Goupillaud, P., A. Grossman, et al. (1984). "Cycle-Octave and Related Transforms in
Seismic Signal Analysis." Geoexploration 23(1): 85-102.
Gusnard, D. A. and M. E. Raichle (2001). "Searching for a baseline: Functional
imaging and the resting human brain." Nat Rev Neurosci 2(10): 685±694.
Harmony, T., T. Fernandez, et al. (1996). "EEG delta activity: an indicator of attention
to internal processing during performance of mental tasks." Int J
Psychophysiol 24(1-2): 161-71.
Jensen, O., J. Gelfand, et al. (2002). "Oscillations in the alpha band (9-12 Hz)
increase with memory load during retention in a short-term memory task."
Cereb Cortex 12(8): 877-82.
Jensen, O. and C. D. Tesche (2002). "Frontal theta activity in humans increases with
memory load in a working memory task." Eur J Neurosci 15(8): 1395-9.
Kappenman, E. S. and S. J. Luck (2010). "The effects of electrode impedance on
data quality and statistical significance in ERP recordings." Psychophysiology
47(5): 888-904.
Klimesch, W. (1999). "EEG alpha and theta oscillations reflect cognitive and memory
performance: a review and analysis." Brain Res Brain Res Rev 29(2-3): 169-
95.
Lang, A. H., O. Eerola, et al. (1995). "Practical issues in the clinical application of
mismatch negativity." Ear Hear 16(1): 118-30.
Laufs, H., J. L. Holt, et al. (2006). "Where the BOLD signal goes when alpha EEG
leaves." Neuroimage 31(4): 1408-18.
Loomis, A. L., E. N. Harvey, et al. (1937). "Cerebral states during sleep, as studied
by human brain potential." Journal of Experimental Psychology 21(2): 127-
144.

!
23!
Lutz, A. and E. Thompson (2003). "Neurophenomenology Integrating Subjective
Experience and Brain Dynamics in the Neuroscience of Consciousness."
Journal of Consciousness Studies 10(9-10): 31-52.
Makeig, S. and M. Inlow (1993). "Lapses in alertness: coherence of fluctuations in
performance and EEG spectrum." Electroencephalogr Clin Neurophysiol
86(1): 23-35.
Makeig, S., M. Westerfield, et al. (2002). "Dynamic brain sources of visual evoked
responses." Science 295(5555): 690-4.
Mantini, D., M. G. Perrucci, et al. (2007). "Electrophysiological signatures of resting
state networks in the human brain." Proc Natl Acad Sci U S A 104(32): 13170-
5.
Maris, E. and R. Oostenveld (2007). "Nonparametric statistical testing of EEG- and
MEG-data." J Neurosci Methods 164(1): 177-90.
Mason, M. F., M. I. Norton, et al. (2007). "Wandering minds: the default network and
stimulus-independent thought." Science 315(5810): 393-5.
Mazoyer, B., L. Zago, et al. (2001). "Cortical networks for working memory and
executive functions sustain the conscious resting state in man." Brain Res Bull
54(3): 287-98.
Naatanen, R., P. Paavilainen, et al. (2007). "The mismatch negativity (MMN) in basic
research of central auditory processing: a review." Clin Neurophysiol 118(12):
2544-90.
Naatanen, R. and T. Picton (1987). "The N1 wave of the human electric and
magnetic response to sound: a review and an analysis of the component
structure." Psychophysiology 24(4): 375-425.
Nittono, H., D. Momose, et al. (2001). "The vanishing point of the mismatch negativity
at sleep onset." Clin Neurophysiol 112(5): 732-9.
Oken, B. S., M. C. Salinsky, et al. (2006). "Vigilance, alertness, or sustained
attention: physiological basis and measurement." Clin Neurophysiol 117(9):
1885-901.
Olbrich, S., C. Mulert, et al. (2009). "EEG-vigilance and BOLD effect during
simultaneous EEG/fMRI measurement." Neuroimage 45(2): 319-32.
Onton, J., A. Delorme, et al. (2005). "Frontal midline EEG dynamics during working
memory." Neuroimage 27(2): 341-56.
Ray, W. J. and H. W. Cole (1985). "EEG alpha activity reflects attentional demands,
and beta activity reflects emotional and cognitive processes." Science
228(4700): 750-2.
Roth, B. (1961). "The clinical and theoretical importance of EEG rhythms
corresponding to states of lowered vigilance." Electroencephalogr Clin
Neurophysiol 13: 395-9.
Sabri, M., S. Labelle, et al. (2003). "Effects of sleep onset on the mismatch negativity
(MMN) to frequency deviants using a rapid rate of presentation." Brain Res
Cogn Brain Res 17(1): 164-76.
Sabri, M., E. Liebenthal, et al. (2006). "Attentional modulation in the detection of
irrelevant deviance: a simultaneous ERP/fMRI study." J Cogn Neurosci 18(5):
689-700.
Sallinen, M. and H. Lyytinen (1997). "Mismatch negativity during objective and
subjective sleepiness." Psychophysiology 34(6): 694-702.
Schooler, J. W. (2002). "Re-representing consciousness: dissociations between
experience and meta-consciousness." Trends in Cognitive Science 6(8): 339-
344.

!
24!
Schreckenberger, M., C. Lange-Asschenfeldt, et al. (2004). "The thalamus as the
generator and modulator of EEG alpha rhythm: a combined PET/EEG study
with lorazepam challenge in humans." Neuroimage 22(2): 637-44.
Smallwood, J., E. Beach, et al. (2008). "Going AWOL in the brain: mind wandering
reduces cortical analysis of external events." J Cogn Neurosci 20(3): 458-69.
Smallwood, J. and J. W. Schooler (2006). "The restless mind." Psychological Bulletin
132(6): 946-58.
Smit, A. S., H. A. Droogleever Fortuyn, et al. (2005). "Diurnal spectral EEG
fluctuations in narcoleptic patients during rest and reaction time tasks." J
Sleep Res 14(4): 455-61.
Sonuga-Barke, E. J. S. and F. X. Castellanos (2007). "Spontaneous attentional
ÀXFWXDWLRQVLn impaired states and pathological conditions: A neurobiological
hypothesis." Neuroscience and Biobehavioral Reviews 31(7): 977-986.
Trungpa, C. (2004). Training the Mind and Cultivating Loving-Kindness. Boston,
Shambhala South Asia Editions.
Wagner, U., S. Gais, et al. (2004). "Sleep inspires insight." Nature 427(6972): 352-5.
Winter, O., A. Kok, et al. (1995). "Auditory event-related potentials to deviant stimuli
during drowsiness and stage 2 sleep." Electroencephalogr Clin Neurophysiol
96(5): 398-412.

!
25!
FIGURES
)LJXUHTime frequency decomposition of transition from mind wandering to breath
focus at electrode site Oz. Mind wandering was defined as the period preceding the
meta-conscious event (button press) and breath focus was defined as the period
following the meta-conscious event. Topographic maps of power difference are
shown for the 2-3.5 Hz (d), 4-7 Hz (q) , 15-30Hz (b) frequency bands from -8 to -2
seconds before and from 2 to 8 seconds after the button press. Topographic map of
differential power is shown for the 9-11Hz (a) frequency band from -6 to -2 seconds
before and 2 to 6 seconds after the button press. (unlike other frequencies bands, the
difference of power in the alpha band between mind wandering and breath focus was
not significant for a larger time interval). Areas of statistical significance (p<0.05) are
highlighted on the topographic maps (shaded areas represent non-significant regions
for a; all electrodes are significant for other frequency bands). The black dot
represents the position of electrode Oz.
)LJXUHEffect of the attentional state on the grand-average ERP for oddball and
standard stimuli. A,B: ERP at electrode site Fz for the mind wandering and the breath
focus state for standard (A) and oddball (B) stimuli. The shaded area surrounding
each curve represents the standard error of the mean. C, D: Topographical maps of
the average ERP difference between mind wandering and breath focus for standard
(C) and oddball (D) stimuli from 180 to 280ms after stimuli presentation
(corresponding to the yellow highlighted region on the ERP plots). Non-significant
areas are grayed out in topographic maps and the black dot indicates the position of
electrode Fz.
)LJXUHMismatch negativity during the mind wandering and breath focus states. A,

!
26!
B: grand-averaged ERP to standard and oddball stimuli for the breath focus and mind
wandering state. As in figure 1, the shaded area surrounding each curve represents
the standard error of the mean. C, D: Topographical difference maps between the
mean ERPs to oddball and standard stimuli (Mismatch negativity MMN) for the
breath focus and the mind wandering state. E: Topographical difference map
between MMN maps in breath focus and mind wandering condition (map C ± map
D). Non-significant areas are grayed out on the topographic maps and the black dot
indicates the position of electrode Fz.
)LJXUHEffect of attentional state and type of stimulus on event related spectral
perturbation. The central panel indicates the grand-average ERSP at electrode site
Oz averaged over both oddball (Odd) and standard (Std) stimuli for both the mind
wandering (MW) and the breath focus (BF) state. Topographical maps of power
difference between the mind wandering and the breath focus conditions are shown
for oddball and standard stimuli at given time-frequency regions of interest (dotted
rectangles). Shaded areas on the topographical maps represent non-significant
regions. Topographical maps for which there were no significant electrodes are
indicated using a blank map ZLWKWKHVLJQ³QV´QRQ-significant). The black dot
represents the position of electrode Oz

3
4
7
10
12
20
25
30
15
5
6
8
9
-8 -6 -4 -2 0246 8
Time (sec)
Mind Wandering Breath Focus
Button press
40
43
47
50
53
10log10(µV2)
δ
θ
α
β
δ
0.9
1.2
dB
0.5 -2.2
-1.4
-0.8
dB
αβ
-0.8
-0.5
-0.3
dB
θ1.2
0.5
0.9
dB

Standard
Oddball
Time (ms)
Potential(μV)
Fz
0 100 200 300 400 500
-6
0
6
0 100 200 300 400 500
-6
0
6
Potential(μV)
breath focus
mind wandering
breath focus
mind wandering dB
2
1
0
Fz
A
B
C
D

0.5
-0.8
dB
-2
0.5
-0.8
dB
-2
C
D
E
0 100 200 300 400
-6
0
6
Breath focus
Fz
standard
oddball
6
0 100 200 300 400
-6
0
Time (ms)
Mind wandering
Fz
oddball
standard
Potential(μV) Potential(μV)
A
B

Frequency (Hz)
−200 0 200 400 600
3
4
7
12
20
33
2
1
0
-1
dB
Oz
Time (ms)
MW - BF
Odd Std
dB
01
0.5
MW - BF
Odd Std
dB
01
0.5
MW - BF
Odd Std
dB
0.4
-0.2
-0.05
ns
MW - BF
Odd Std
ns ns
ns