ArticlePDF Available

Detection of mind wandering using EEG: Within and across individuals


Abstract and Figures

Mind wandering is often characterized by attention oriented away from an external task towards our internal, self-generated thoughts. This universal phenomenon has been linked to numerous disruptive functional outcomes, including performance errors and negative affect. Despite its prevalence and impact, studies to date have yet to identify robust behavioral signatures, making unobtrusive, yet reliable detection of mind wandering a difficult but important task for future applications. Here we examined whether electrophysiological measures can be used in machine learning models to accurately predict mind wandering states. We recorded scalp EEG from participants as they performed an auditory target detection task and self-reported whether they were on task or mind wandering. We successfully classified attention states both within (person-dependent) and across (person-independent) individuals using event-related potential (ERP) measures. Non-linear and linear machine learning models detected mind wandering above-chance within subjects: support vector machine (AUC = 0.715) and logistic regression (AUC = 0.635). Importantly, these models also generalized across subjects: support vector machine (AUC = 0.613) and logistic regression (AUC = 0.609), suggesting we can reliably predict a given individual’s attention state based on ERP patterns observed in the group. This study is the first to demonstrate that machine learning models can generalize to “never-seen-before” individuals using electrophysiological measures, highlighting their potential for real-time prediction of covert attention states.
Content may be subject to copyright.
Detection of mind wandering using EEG:
Within and across individuals
Henry W. Dong
, Caitlin Mills
, Robert T. Knight
, Julia W. Y. Kam
1Helen Wills Neuroscience Institute, University of California, Berkeley, Berkeley, California, UnitedStates of
America, 2Department of Psychology, University of New Hampshire, Durham, New Hampshire, United
States of America, 3Department of Psychology, University of Calgary, Calgary, Alberta, Canada,
4Hotchkiss Brain Institute, Calgary, Alberta, Canada
Mind wandering is often characterized by attention oriented away from an external task
towards our internal, self-generated thoughts. This universal phenomenon has been linked
to numerous disruptive functional outcomes, including performance errors and negative
affect. Despite its prevalence and impact, studies to date have yet to identify robust behav-
ioral signatures, making unobtrusive, yet reliable detection of mind wandering a difficult but
important task for future applications. Here we examined whether electrophysiological mea-
sures can be used in machine learning models to accurately predict mind wandering states.
We recorded scalp EEG from participants as they performed an auditory target detection
task and self-reported whether they were on task or mind wandering. We successfully clas-
sified attention states both within (person-dependent) and across (person-independent)
individuals using event-related potential (ERP) measures. Non-linear and linear machine
learning models detected mind wandering above-chance within subjects: support vector
machine (AUC = 0.715) and logistic regression (AUC = 0.635). Importantly, these models
also generalized across subjects: support vector machine (AUC = 0.613) and logistic regres-
sion (AUC = 0.609), suggesting we can reliably predict a given individual’s attention state
based on ERP patterns observed in the group. This study is the first to demonstrate that
machine learning models can generalize to “never-seen-before” individuals using
electrophysiological measures, highlighting their potential for real-time prediction of covert
attention states.
We frequently find ourselves drifting away from a conversation, our work, or the movie in
front of us, towards our inner world. Often referred to as mind wandering, this phenomenon
has been traditionally characterized as an attentional shift from externally oriented task-related
thoughts to internally oriented task-unrelated thoughts [13]. More recently, the specific defi-
nition of mind wandering is under debate [4,5]. Mind wandering has been associated with
numerous benefits, including future planning, creativity, and problem solving [68]. However,
previous studies have also established robust negative associations with mind wandering:
PLOS ONE | May 12, 2021 1 / 18
Citation: Dong HW, Mills C, Knight RT, Kam JWY
(2021) Detection of mind wandering using EEG:
Within and across individuals. PLoS ONE 16(5):
Editor: Friedhelm Schwenker, Ulm University,
Received: August 8, 2020
Accepted: April 27, 2021
Published: May 12, 2021
Copyright: ©2021 Dong et al. This is an open
access article distributed under the terms of the
Creative Commons Attribution License, which
permits unrestricted use, distribution, and
reproduction in any medium, provided the original
author and source are credited.
Data Availability Statement: Data included in this
study are available on the Open Science
Framework (OSF) repository. URL:
be573fc433064165aee8019f2891b3fe DOI: 10.
Funding: This work was supported by National
Institute of Neurological Disorders and Stroke
(NINDS; R3721135
and National Institute of Mental Health (NIMH; grant PO1
MH109420-01 for RTK, and Natural Sciences and
including higher levels of negative affect [9,10] and impaired performance in a variety of
externally oriented tasks, such as target detection [11], performance monitoring [12], and
reading comprehension [1315]. Given the potential negative impact of mind wandering on
performance, reliable detection of this phenomenon provides a step towards optimizing task
performance in daily life.
Mind wandering is an inherently internal process that often occurs in the absence of overt
behavioral markers, making it difficult to detect and combat in real-time. A unique barrier for
mind wandering research is its current overreliance on self-reported experience. In particular,
the most commonly used and well-validated approach to study mind wandering is the online
thought sampling method [2,16], in which subjects are occasionally interrupted throughout
an externally oriented task and asked to indicate whether they were having a task-related
thought (i.e. on task) or task-unrelated thought (i.e. mind wandering). One of the main advan-
tages of thought sampling is that it provides a direct, in-the-moment measure of one’s atten-
tional state. However, this approach may be impacted by demand characteristics or lack of
awareness of attention state [16,17]. Thought sampling may also change the nature of the task
itself since it requires constant interruptions. Unobtrusive detection of mind wandering using
machine learning methods thus offers a potential solution that overcomes these challenges and
provides avenues for applications that can address the negative impacts of mind wandering in
real-time [18]. Establishing the validation and effectiveness of machine learning in detecting
mind wandering across contexts has the potential to eventually replace the need for thought
sampling to determine the occurrence of mind wandering.
Previous successful attempts of mind wandering detection primarily used behavioral mea-
sures such as eye tracking and pupillometry [1922] or task-related measures, such as driving
performance [23,24] and reading time [25]. Studies have also used physiological measures
such as heart rate and skin conductance [26] as well as synchronization between respiration
and sensory pressure [27]. These findings serve to highlight the value of using behavioral and
physiological measures to detect mind wandering at above chance levels. Compared to these
indirect markers, neural measures may be more effective at directly capturing this inherently
covert attention state. Here, we assessed the utility of a neural measure for mind wandering
detection by examining electrical activity originating from the brain using scalp EEG during
this internally oriented state.
Electrophysiological markers of mind wandering
Ample evidence from scalp EEG studies has established a distinct set of electrophysiological
signatures of mind wandering, which is promising for real-time detection [2834]. Given that
scalp EEG is comparably low cost, and can be implemented outside of the laboratory, these
EEG measures could be ideal for real-time detection of this phenomenon in the real world. In
particular, the P1 and N1 ERP components in response to visual and auditory probes in a tar-
get detection task are reduced during mind wandering [33,35,36]. This indicates that sen-
sory-evoked responses to both visual and auditory inputs were significantly attenuated,
suggesting that mind wandering disrupts external perception regardless of sensory modality.
Similarly, several studies have documented a reduction of the P3 ERP component during peri-
ods of mind wandering relative to on task [12,34,37,38], demonstrating an overall attenua-
tion of higher-level cognitive processes. Together, these findings suggest that mind wandering
periods are associated with reduced external processing as observed in changes in ERP ampli-
tude. These studies are consistent with the executive function model of mind wandering,
which posits that in order to facilitate mind wandering, our executive resources are disengaged
from the external task and instead directed internally to our thoughts [2,7].
Detection of mind wandering using EEG
PLOS ONE | May 12, 2021 2 / 18
Engineering Research Council (NSERC; https:// for JWYK.
The funders had no role in study design, data
collection and analysis, decision to publish, or
preparation of the manuscript.
Competing interests: The authors have declared
that no competing interests exist.
Prior EEG studies have also reported changes in low frequency power during mind wander-
ing. Specifically, Braboszcz and Delorme reported increases in theta activity and decreases in
alpha activity during mind wandering [28]. Using a similar experimental design, van Son and
colleagues [39] found a higher ratio of theta and beta activity in the frontal cortex during mind
wandering. In contrast to these findings, others have reported increased frontocentral theta
power during external cognition [4042], whereas increased posterior alpha power has been
implicated in mind wandering [23,38]. These variable patterns in low frequency power as a
marker of mind wandering may be driven by differences in stimulus modality, experimental
manipulation of top-down processes, or the electrode sites at which power was measured.
Importantly, given the overlapping information between low frequency power and ERP com-
ponents (typically measured between 1-30Hz), and that ERP patterns are relatively more con-
sistent across studies, the current study used ERP measures as features in the machine learning
Using EEG for mind wandering detection
Converging evidence points towards several reliable EEG correlates of mind wandering, and
several studies to date have attempted to build detectors of mind wandering based solely on
EEG measures. Kawashima and Kumano [43] used EEG signals (power and coherence in
delta, theta, alpha, beta, and gamma frequency bands) to predict mind wandering intensity
during a sustained attention to response task. They found that non-linear models using multi-
ple electrodes resulted in higher prediction accuracy of mind wandering intensity than linear
models using a single electrode. Jin and colleagues [44] extended this work by predicting mind
wandering with a nonlinear model that generalized across tasks within individual subjects.
Specifically, they used a support vector machine to predict mind wandering with EEG markers
(including the P1, N1, and P3 ERP components, as well as theta and alpha power and coher-
ence), reporting an average on task/off task classification accuracy of 60% that generalized
across two visual tasks. This task generalization is noteworthy as it suggests that models trained
on EEG markers may detect mind-wandering without needing to first train on new tasks. This
group [45] also demonstrated that mind wandering is not dependent on and is quantitatively
different from subject vigilance and current task demands, by reporting that a classifier trained
through thought sampling outperformed classifiers trained through either vigilance or task
demands. Dhindsa and colleagues [46] extended these findings by detecting mind wandering
in real world settings. They recorded EEG activity during live lectures and used frequency
band power measurements (theta, alpha, and beta) to achieve an average detection accuracy of
80–83%. Finally, Tasika and colleagues [47] employed a multi-step framework that leverages a
J48 decision tree classifier and a support vector machine with a radial basis function kernel to
detect mind wandering using EEG, and reported an average accuracy as high as 84.49% based
on two individuals.
An important property that is not clearly addressed in previous work, however, is the gener-
alizability of models across participants: a model that identifies an optimal set of features at the
group level that can accurately predict attention states of another individual not in that group
(i.e. person-independent). Although this approach typically results in overall lower classifica-
tion performance, it allows for more flexible generalizability for real-time prediction. Previous
attempts using machine learning and EEG measures exclusively to detect mind wandering
used data from the same individuals to train and test the models. In other words, there was
dependence within individuals, which means the models likely do not generalize well to new
subjects. We aim to improve on mind wandering detection using only EEG by building models
that generalize across individuals in a person-independent manner.
Detection of mind wandering using EEG
PLOS ONE | May 12, 2021 3 / 18
This proof-of-concept study examined whether machine learning models using EEG mea-
sures can detect mind wandering 1) within individuals and 2) across individuals. Using
thought sampling during a target detection task, we asked subjects to report their attention
state throughout the task as we recorded their EEG. Our study included EEG markers, specifi-
cally ERP components that have previously demonstrated reliable attentional differences, as
features in two machine learning models. To our knowledge, this is the first study to establish
that machine learning using EEG features is capable of detecting mind wandering both within
individuals (i.e. person-dependent) and across individuals (i.e. person-independent).
To address this issue, we asked participants to perform an auditory target detection task
while their EEG is being recorded. In order to obtain a measure of participants’ attention state
in the moment, participants were asked to report their attentional state as on task or mind
wandering at pseudorandom intervals throughout the task. We extracted EEG measures of
interest (namely N1 and P3 ERP) as a function of the reported attention states. Using these
EEG measures, we built linear and non-linear machine learning models to determine whether
these measures can be used to classify attention states both within and across individuals, and
if so, which models led to superior performance. For classification within subjects, we used
each subject’s own EEG markers of attention state for prediction, which allowed us to maxi-
mize prediction accuracy for that individual. For classification across subjects, our models
trained on one set of data and attempted to identify an optimal algorithm that can predict
attentional state of individuals not part of the training group data. Together, these two
approaches help determine whether machine learning with EEG measures can accurately pre-
dict mind wandering within and across individuals.
Fourteen subjects participated in the experiment (9 females, 5 males; age range: 24–75, M±
SD = 51.9±14.7). Although the sample size appears small, it is sufficient for addressing our pri-
mary aim of mind wandering detection. Specifically, it is comparable with previous studies
using within-subject training and testing sets that included fewer than 20 subjects to detect
mind wandering using EEG features (i.e. features that were derived from EEG data) [44,46].
Further, if we can predict mind wandering with this sample size for the across-subject analyses,
this finding provides a lower threshold necessary for accurately detecting mind wandering
using EEG features. All subjects provided informed written consent and were reimbursed for
their participation. This study was approved by the Institutional Review Board at the Univer-
sity of California, Berkeley.
Task stimuli and paradigm
Subjects sat in a dark room and performed an auditory target detection task [36]. They were
presented with a series of standard tones (800Hz) and target tones (1000Hz) in a random
order with probabilities of 0.8 and 0.2, respectively. There was a total of 1500 tones, 1200 of
which were standard tones and 300 of which were target tones. Each sound was a pure tone
presented at 75 dB SL that lasted 200ms, and the inter-trial interval was randomly jittered
between 800-1200ms. Subjects were instructed to press a button to target tones as quickly and
accurately as possible. They were asked to keep their eyes fixated on the cross in the center of
the screen at all times. Reaction time to the target tones was recorded, and accuracy was con-
sidered across both tones (i.e. detection of target tones and correct rejection of standard
tones). Attentional differences in behavioral measures were assessed using dependent samples
Detection of mind wandering using EEG
PLOS ONE | May 12, 2021 4 / 18
During the task, we occasionally presented thought probes that asked subjects to report
their attention state as either “on task” or “mind wandering.” To ensure subjects understood
the meaning of these attention states, we provided them with clear and detailed definitions
and examples. “On task” was defined as one’s attention being firmly directed towards the tar-
get detection task, and “mind wandering” was defined as one’s attention being oriented away
from the task. For each thought probe, we extracted the 10 trials (approximately 15 seconds)
prior to the probe, and labeled these 10 trials according to the subject’s response to the thought
probe. For instance, if the response to one thought probe was “mind wandering”, then the 10
trials preceding that thought probe received a label of mind wandering, and are referred to as
mind wandering trials thereafter. This time window has been previously used in ERP studies
[12,33,37,48] in order to maximize the number of trials that can be included to create a reli-
able ERP average while still maintaining a reasonable validity of the attentional report. Earlier
studies have also provided detailed justification for using this time window [33,34]. Each
block ended with a thought probe. There was a total of 25 blocks, with varying block duration
(i.e. 45 to 75 seconds) to prevent subjects from anticipating thought probe occurrence. There-
fore, a maximum number of 250 trials (10 trials x 25 blocks) were included in subsequent anal-
yses for each subject.
Despite the shortcomings of the thought sampling approach, it serves to provide the
labelled attention reports for our supervised machine learning models (as described below).
Therefore, thought sampling remains a valuable tool for validating machine learning models
in mind wandering. Once sufficient evidence accumulates that validate machine learning
models across different experimental paradigms, we can strive to reduce reliance on this mea-
sure in future studies.
EEG acquisition and preprocessing
EEG was recorded reference-free continuously from 64 active electrodes mounted on a cap
using the Biosemi ActiveTwo system ( Electrodes were placed according
to the International 10–20 system. Continuous EEG data were amplified and digitized at 1024
Hz, and bandpass filtered between 1Hz and 50Hz. Vertical and horizontal eye movements
were recorded from electrodes above and below the right eye and two electrodes placed at the
right and left outer canthus. EEG data were down-sampled to 512 Hz, then high-pass filtered
at 1 Hz (as this is ideal for independent component analysis [49]), and notch filtered at 60 Hz.
Electrodes with excessively noisy signals were removed and replaced with an interpolation
from neighboring electrodes using spherical spline interpolation [50]. Independent compo-
nent analysis was used to detect and remove eye movement and muscle artifacts. Continuous
EEG data were then segmented into 3000ms epochs, beginning at 1000ms prior to stimulus
onset. Each trial was visually inspected for any remaining artifacts, which were further manu-
ally removed. Common average reference was then applied to each subject’s data prior to ERP
analysis. EEG data pre-processing and analysis were adapted from Kam and colleagues [51]
and performed using EEGLAB [52,53] and custom Matlab scripts.
ERP analysis
EEG signals were bandpass filtered at 1-15Hz for ERP analysis [5456]. All ERPs were quanti-
fied by the peak amplitude measure relative to a -200 to 0 pre-stimulus baseline. For N1, we
extracted the minimum amplitude in the 80–120 ms post-stimulus time window over fronto-
central midline sites (FC1, FCz, FC2), where N1 is typically maximal and also maximal in our
data. For P3, we extracted the maximum amplitude in the 400–600 ms post-stimulus time win-
dow over parietal sites (P1, Pz, P2), where P3 is typically maximal and also maximal in our data.
Detection of mind wandering using EEG
PLOS ONE | May 12, 2021 5 / 18
For each ERP measure, we averaged across the channels and time points of interest as
described above. We considered up to the 10 trials (excluding artifactual trials) preceding the
thought probe at the end of each block, and categorized them according to the reported atten-
tion state. In order to ensure our data are consistent with previous findings, we first imple-
mented univariate analyses to examine attentional effects in the N1 and P3 ERP components.
In particular, we statistically compared the peak ERP amplitudes between on task and mind
wandering states by implementing repeated measures ANOVAs for each ERP component (i.e.
N1 and P3) to examine attentional effects (i.e. on task and mind wandering) while taking into
account tone differences (i.e. standard and target tones). Paired samples t-tests were conducted
post hoc to examine attentional effects separately for each tone. These analyses serve the pur-
pose of ensuring that ERP patterns across on task and mind wandering states are consistent
with previous studies.
For machine learning analyses, we extracted information at the block level. Specifically, we
computed the mean and standard deviation across up to 10 trials for each block in order to
obtain a stable ERP measure (via averaging across trials per block) and have sufficient data
points per subject to input into the machine learning model (via deriving one input per block).
The alternative options are suboptimal, including measuring ERPs at the single trial level
which results in unreliable estimates of ERP, or averaging across all blocks to yield one grand
average value per subject which results in insufficient inputs for machine learning models
to classify attention states. Given the temporal fluctuations of our attention state, our
approach allows us to account for the ebb and flow of attention throughout the task and its
corresponding changes in the electrophysiological measures across time within individuals.
Accordingly, for each block, we extracted the following features: 1) the mean across the 10 pre-
ceding trials for each ERP measure (the N1 minimum amplitude and the P3 maximum ampli-
tude), as an index of the magnitude of electrophysiological response, and 2) the standard
deviation across the 10 trials for each ERP measure, as an index of the variability in our
response to external events. In essence, each subject will have up to as many inputs as blocks
completed (i.e. 25) for each ERP component (i.e. N1 and P3) and descriptive measure (mean
and standard deviation of ERP peak amplitudes across 10 trials per block) into the machine
learning models.
Statistical analysis
To predict attention states, we included the aforementioned ERP measures as features into a
machine learning model that makes a binary classification: on task (0) or mind wandering (1).
In particular, we derived the mean and standard deviation of the N1 and P3 peak amplitudes
of the last 10 trials per block. For these analyses, we focused on ERP amplitudes in response to
standard tones only since there were very few target tones (i.e. on average two) occurring
within the last 10 trials within a block. This low number of target tones not only yields an unre-
liable estimate of the ERP response to target tones, but it could reduce generalization across
individuals. Using Scikit-learn (version 0.20.2), we built two machine learning models to dis-
criminate between the two attention states, including one linear model (i.e. linear logistic
regression) and one non-linear model (i.e. a support vector machine (SVM) with a radial basis
function (RBF) kernel). We computed the mean amplitude across all blocks in the on task
state and subtracted this mean from the data for both attention states; this additional step nor-
malizes the data, which makes the two conditions more comparable across different subjects
[57]. This normalization was performed independently for each feature and each subject. We
adopted this approach so that our machine learning models would likely be less impacted by
individual differences in the absolute values of each EEG measure.
Detection of mind wandering using EEG
PLOS ONE | May 12, 2021 6 / 18
Class imbalance. The number of data points in the two attention state conditions was
slightly imbalanced (n= 156 for on task, and n= 193 for mind wandering). Class imbalance
often poses a challenge for supervised classifiers due to more exposure to the majority class in
the training data. Similar to many other eye gaze-based mind wandering detectors [21,25,58,
59], we corrected for this imbalance prior to training the models using an oversampling tech-
nique to increase the number of data points in the minority class in the training data. Specifi-
cally, we used Synthetic Minority Over-Sampling Technique (SMOTE) [60], which creates
new “synthetic” instances of on task (the minority class) to balance the classes in the training
set; the model is therefore trained on equal numbers of both classes in order to better learn the
patterns. SMOTE creates the synthetic on-task instances through a linear interpolation of the
feature values from a series of “real” nearest-neighbor values in the data. SMOTE is a com-
monly used approach for mind wandering classification with machine learning, given that the
classes are typically imbalanced [59,61]. One subject was excluded from subsequent analyses
because they did not report enough instances of the minority class (N <5) to generate syn-
thetic data points using the SMOTE technique. Although it is also possible to remove data
points from the mind wandering condition with more data points to achieve a class balance,
we implemented the SMOTE in order to maximize the amount of data that can be used for
training in the models.
The models were used to classify attention states in two ways 1) within subjects (i.e. person-
dependent models), the current gold standard in the literature using EEG features, and 2)
across subjects (i.e. person-independent models). Classification within subjects attempts to
detect mind wandering on an individual basis, using each subject’s own electrophysiological
signatures of attention state for prediction. This approach is useful for maximizing prediction
accuracy for that particular individual, but the model does not generalize well to new individu-
als. In contrast, classification across subjects attempts to accurately classify attention states
in “never-before seen” subjects for greater generalizability. These models attempt to find an
optimal algorithm that can predict attentional states of individuals who were not part of the
training group data. This method has been successfully implemented using behavioral and
pupillometry measures [19,25]; however, no studies to our knowledge have successfully classi-
fied attention states using a generalizable model across subjects with EEG features.
Cross-validation methods. To classify on task vs. mind wandering within a subject (i.e.
person-dependent models), we used k-fold cross validation, with 5 folds on 25 instances
(which represent the 25 blocks of data) for each subject. In order to ensure class balance within
subjects for the classification analysis, SMOTE was performed on the training set within each
For training the models across subjects (i.e. person-independent models) we used leave-
one-subject-out cross-validation, which is similar to k-fold cross validation, with the exception
that the training and testing sets are completely independent. Using this technique, one subject
was reserved as the testing set, and the remaining k-1 subjects were used as the training set.
This was repeated k times, where k = the number of subjects, such that each subject was used
as the testing set once. This validation was performed to ensure that the training and testing
sets are both exclusive and independent, and that the model generalizes across new subjects.
We chose this validation technique, as opposed to the traditional k-fold cross validation, due
to our limited sample size. Importantly, it closely emulates real-life applications of mind wan-
dering detection, where a model can be previously trained on a set of data gathered from mul-
tiple subjects, and then tested on a new individual.
Model evaluation. Model performance was evaluated using three common metrics: accu-
racy, area under the curve (AUC), and Matthews Correlation Coefficient (MCC). Accuracy is
the number of correct predictions of attention state over the number of total predictions, and
Detection of mind wandering using EEG
PLOS ONE | May 12, 2021 7 / 18
ranges from 0 (no predictions correct) to 1 (all predictions correct). Notably, accuracy by itself
is somewhat problematic when class imbalance exists, since prediction of the majority class
will be inherently above 50% by chance. We therefore focus on the other two metrics that are
not sensitive to class imbalance.
AUC is one of the most widely used metrics for binary classification. It is equivalent to the
probability that the model will rank a randomly chosen positive example (e.g. mind wander-
ing) higher than a randomly chosen negative example (e.g. on task). This measure can be rep-
resented as the area under the curve of a plot of the false positive rate vs. the true positive rate.
AUC ranges from 0 (no predictions correct) to 1 (all predictions correct), with chance level at
0.5. MCC takes into account true and false positives and negatives, and outputs a value that
ranges from -1 (no predictions correct) to 1 (all predictions correct), with chance level at 0.
Together, these metrics demonstrate whether the models are capable of successful classifica-
tion of attention states, and assess classification performance with respect to chance and imbal-
anced classes.
We additionally evaluated confusion matrices of the best performing models, which
allowed us to visualize the performance of each model and summarize the true and false posi-
tives and negatives. The confusion matrix reveals what type of errors the model makes: specifi-
cally, whether the model shows any bias or skew towards a particular class, and if so, how. It is
possible to examine the true positive rate (sensitivity) and true negative rate (specificity) to
determine which of the two models result in more accurate prediction. Similarly, it is possible
to examine the false positive and false negative rates to determine if the model leans towards
any decision-making errors. For example, a high sensitivity and high false positive rate may
point to over-classification of the positive instance.
Behavioral performance
Subjects reported mind wandering 55% of the time (range: 20%– 88%, SE = 5.3%) and being
on task 45% of the time. This is consistent with the typical breakdown of self-reported atten-
tion states in the literature for this type of task [34,35,37]. Mean reaction time and accuracy
are shown in Fig 1. There was a trend for slower response time during mind wandering
(M= 535 ms, SE = 14 ms) compared to on task periods (M= 511 ms, SE = 14 ms; t(12) = -1.94,
p= 0.076). However, accuracy did not differ between attention states for mind wandering
(M= 0.9619, SE = 0.030) compared to on task (M= 0.9623, SE = 0.028; t(12) = 0.098,
p= 0.924).
Univariate analyses on ERP components
In order to check the veracity of the attentional reports, we first evaluated whether our ERP
measures reveal attentional differences that are consistent with previous studies. Specifically, if
our ERP measures show reduced amplitudes during mind wandering, this would provide cor-
roborating evidence for the attentional reports. To assess conditional differences in ERP mea-
sures, we implemented repeated measures ANOVA with attention (on task and mind
wandering) and tone (standard and target) as within subject factors, separately for the N1 and
P3 ERP components. These results are reported in Table 1.
As expected, the main effect of tone was significant for both the N1 and P3, with increased
amplitude during target tones relative to standard tones. For N1, there was a near significant
main attentional effect, characterized by reduced N1 during mind wandering relative to on task
periods. Although the attention x tone interaction was not significant, we implemented follow
up analyses using paired samples t-tests to examine whether attentional effects varied for each
Detection of mind wandering using EEG
PLOS ONE | May 12, 2021 8 / 18
tone as was observed in Fig 2. Consistent with previous findings of attentional effects in sensory
ERP components [33,36], our analyses revealed a significant attentional effect for the N1 in
response to target tones (t(13) = -2.83, p= 0.014) but not standard tones (t(13) = -0.72, p =
0.485). Neither the main attention effect nor interaction effect were significant for the P3.
Machine learning
Person-dependent classification performance. Individual accuracy for subjects with the
non-linear SVM with RBF kernel ranged from 0.383 to 0.960 (M= 0.643, SE = 0.044), whereas
accuracy for the linear logistic regression model ranged from 0.375 to 0.843 (M= 0.621,
SE = 0.038). The range of accuracy values observed in our study, as well as the variability in
accuracy rates across individuals, are similar to accuracy values reported in prior work using
EEG measures for predicting mind wandering [44,46]. Although accuracy might be useful for
comparing findings across different studies, this metric can be biased with class imbalance (as
previously mentioned); therefore, it is more informative to examine the AUC and MCC values
for individual subjects.
Both AUC and MCC metrics also demonstrate, on average, above-chance performance for
the SVM and logistic regression models. AUC ranged from 0.452 to 0.995 (M= 0.715,
SE = 0.0457) for the SVM and 0.291 to 0.858 (M= 0.635, SE = 0.0491) for the logistic regres-
sion model. MCC ranged from -0.162 to 0.919 (M= 0.289, SE = 0.0792) for the SVM and
-0.287 to 0.694 (M= 0.210, SE = 0.0758) for the logistic regression model. Machine learning
performance as indexed by AUC and MCC for each individual subject is shown in Fig 3.
Response Time (ms)
Accuracy (proportion)
Attention State
Attention State
Fig 1. Behavioral results as a function of attention state. (A) Mean response time was slower during mind
wandering (p = 0.076). (B) No difference was observed in accuracy between the two attention states. Error
bars = standard error of the mean; OT = on task; MW = mind wandering.
Table 1. ANOVAs on ERP components.
Features Attention (OT vs. MW) Tone (Standard vs. Target) Attention x Tone Interaction
N1 Min F(1,52) = 3.57, p= 0.064 F(1,52) = 6.23, p= 0.013 F(1,52) = 1.51, p= 0.224
P3 Max F(1,52) = 1.79, p= 0.187 F(1,52) = 58.99, p<.001 F(1,52) = 1.24, p= 0.271
Note: MW = mind wandering. OT = on task.
Repeated measures ANOVAs of main effects of attention and tone as well as attention ×tone interaction, reported separately for N1 and P3 ERP components.
Detection of mind wandering using EEG
PLOS ONE | May 12, 2021 9 / 18
Person-independent classification performance. In this method of classification, we
aimed to generalize the model across subjects by using leave-one-subject-out cross-validation.
Both the nonlinear SVM and logistic regression models performed above chance (MCC >0,
AUC >0.5). Accuracy ranged from 0.514 to 0.750 (M= 0.591, SD = 0.070) for the SVM, and
from 0.461 to 0.752 (M= 0.588, SD = 0.093) for the logistic regression model. MCC values ran-
ged from –0.026 to 0.446 (M= 0.206, SD = 0.152) for the SVM and –0.056 to 0.449 (M= 0.196,
SD = 0.169) for the logistic regression. Finally, AUC values ranged from 0.485 to 0.731
(M= 0.613, SD = 0.085) for the SVM and 0.468 to 0.739 (M= 0.609, SD = 0.096) for the logistic
regression model. Overall, the nonlinear SVM showed the best performance, as indicated by
higher scores in accuracy, AUC, and MCC, as compared to logistic regression. Although there
Time (s)
-0.2 0 0.2 0.4 0.6
Amplitude (uV)
Time (s)
-0.2 0 0.2 0.4 0.6
Amplitude (uV)
OT standard
OT target
MW standard
MW target
Fig 2. Grand average ERP waveforms. N1 was averaged across FC1, FCz and FC2 (left panel), whereas P3 was
averaged across P1, Pz and P2 (right panel). Univariate analyses indicate reduced N1 in response to target tones during
MW relative to OT periods. OT = on task, MW = mind wandering.
Detection of mind wandering using EEG
PLOS ONE | May 12, 2021 10 / 18
is clear variability in the evaluation metrics across participants (as is common for other real-
time detectors [46]), we suggest the average model performance faired quite well for most indi-
viduals. A summary of the results for each model are reported in Table 2.
The ultimate goal of a mind wandering detector is to assess the phenomenon in real time
for measurement and interventions; thus precision (i.e. accurately classifying instances of
mind wandering as mind wandering) is a priority of the model. We therefore evaluated the
confusion matrices of the best models to determine the type of errors the models made (as
shown in Table 3). Neither model exhibited any skew towards sensitivity (i.e. true positives for
predicting mind wandering) or specificity (i.e. true negatives or correct rejections). Taken
AUC Score
MCC Score
12 13
12345678910 11 12 13
SVM Logistic Regression
SVM Logistic Regression
Fig 3. Model performance per subject, as measured by AUC and MCC. AUC performance for each subject for both
models (SVM and logistic regression) is shown in top panel; chance is noted by the black horizontal line at 0.5. MCC
performance for each subject is shown in bottom panel; chance is noted by the black horizontal line at 0. AUC = area
under the curve; MCC = Matthews correlation coefficient; SVM = support vector machine.
Table 2. Model evaluation of person-independent classification performance.
Models Performance Metrics
Accuracy AUC MCC
SVM with RBF Kernel 0.591 (SD = 0.070) 0.613 (SD = 0.085) 0.206 (SD = 0.152)
Logistic Regression 0.588 (SD = 0.093) 0.609 (SD = 0.096) 0.196 (SD = 0.169)
Note: Classification performance indices, including accuracy, AUC, and MCC, are reported for both machine learning models: SVM with RBF kernel and logistic
regression. AUC = area under the curve; MCC = Matthews correlation coefficient; SVM = support vector machine; RBF = radial basis function.
Detection of mind wandering using EEG
PLOS ONE | May 12, 2021 11 / 18
together, the SVM model outperformed the logistic regression model. Not only did the SVM
model show higher sensitivity and specificity, it also showed higher values for AUC and MCC,
which are metrics that are robust against the presence of imbalanced classes.
Feature analysis. To further examine each feature individually, we created models that
were trained on only one feature at a time (Table 4). Given that SVM outperformed logistic
regression, we implemented these analyses with the SVM model only. This post-hoc analysis
allowed us to identify the feature that is most effective in predicting attention states. Our
results indicate that most features performed slightly above chance, with the peak P3 ampli-
tude showing the best performance. Notably, none of the features individually performed as
well as the complete model.
Mind wandering is an intrinsically covert state that is consistently linked to negative affect and
performance decrements. Although it is traditionally measured via self-reports, our proof-of-
concept study attempts to overcome the shortcomings of this approach by developing machine
learning models that can reliably predict its occurrence using EEG measures that directly and
objectively capture neural activity. Such models will be crucial to the development of applica-
tions that can mitigate the negative effects of mind wandering in real-time [18], as well as
future research that will no longer need to rely on constant task interruptions to determine
when mind wandering occurs using thought sampling.
During a target detection task, our univariate analyses demonstrated significant attentional
differences in the sensory ERP components. Our machine learning classifiers using both the
support vector machine and logistic regression models were able to classify mind wandering at
above-chance levels within subjects. Notably, we improved on prior work in this area by
Table 3. Confusion matrices of person-independent models.
Actual MW Actual Not MW
SVM with RBF Kernel Pred. MW 0.570 0.390
Pred. Not MW 0.430 0.610
Logistic Regression Pred. MW 0.528 0.356
Pred. Not MW 0.472 0.644
Note: Confusion matrix for each of the machine learning models: SVM with RBF kernel and logistic regression. Pred.
= predicted; MW = mind wandering; SVM = support vector machine; RBF = radial basis function
Table 4. Model performance for individual features.
Features Performance Metrics
Accuracy AUC MCC
N1 Min 0.489 0.521 0.033
N1 SD 0.322 0.471 -0.084
P3 Max 0.562 0.594 0.171
P3 SD 0.518 0.567 0.106
Note: Model performance metrics, including accuracy, AUC, and MCC, implemented separately for each individual
feature of ERP components. Models for all four features were built with the SVM with RBF kernel. SD = standard
deviation. AUC = area under the curve; MCC = Matthews correlation coefficient; SVM = support vector machine;
RBF = radial basis function.
Detection of mind wandering using EEG
PLOS ONE | May 12, 2021 12 / 18
developing person-independent models that can detect mind wandering using EEG features
without having any prior information about that person. The performance of our person-inde-
pendent models is modest, but comparable to previous studies showing generalizability across
subjects using behavioral measures, eye gaze, and pupillometry [25,58,61]. These findings
underscore, for the first time, the potential for generalizable machine learning models that can
classify mind-wandering states in real-time using electrophysiological measures.
Using machine learning methods with EEG measures to predict mind wandering, we had
better success with the SVM model relative to the logistic regression model in both within-
and across-subject classification of mind wandering states. This finding is consistent with pre-
vious EEG studies that demonstrate the effectiveness of nonlinear models in determining the
boundary between attention states [43,44,46], highlighting the utility of nonlinear models in
classification accuracy and generalizability. Examination of the confusion matrix revealed that
the models were not biased in their predictions, with both models showing negligible differ-
ence between sensitivity and specificity. These unbiased results may be due to the oversam-
pling technique, SMOTE, which addressed the issue of class imbalance between attention
states. The effectiveness of SMOTE has been demonstrated in other work using behavioral
measures in detecting mind wandering [21,59]. A previous study that classified mind wander-
ing states using EEG has reported issues with highly disparate sensitivity and specificity using
other techniques to balance class sizes [44]. These disparities could be attributed to either the
method used to achieve balanced classes (such as SMOTE) or the accuracy of subjects’
responses during thought sampling. Although this is a well-validated approach in measuring
one’s attentional state [2,62], thought sampling does nevertheless rely on self-reports and
therefore depend on accurate and honest responses from subjects. Any bias in subjects’
responses could reduce classification accuracy and potentially result in the observed difference
between sensitivity and specificity. This limitation may explain parts of our data in which
model performance was below chance for at least one subject in the person-dependent model,
suggesting that machine learning models are effective in general in predicting attention states
but not necessarily for every single individual.
Finally, we examined each feature individually by creating separate SVM models for each
feature, in order to evaluate its relative importance in classification accuracy. Interestingly, the
peak P3 amplitude resulted in the highest accuracy, AUC, and MCC, despite not showing sig-
nificant attentional differences in the univariate analyses. These contrasting results may be
puzzling at first glance; however, the differing nature of the averaged ERP amplitudes entered
into the univariate analyses and machine learning models may potentially account for this dif-
ference. While the univariate analyses involved the peak amplitudes averaged across the entire
task, the machine learning models received the peak amplitudes averaged within each of the 25
blocks as input; therefore, it takes into account fluctuations in the P3 amplitude throughout
the task. That peak P3 amplitude contributed to classification accuracy but did not show atten-
tional differences in univariate analyses suggests that statistical significance at the univariate
level captures different information compared to machine learning classifiers. Future work
can clarify the relationship between these types of input data and their corresponding analyses.
These findings underscore the value of considering multivariate patterns of data via machine
learning models in classifying temporally fluctuating attentional states.
Importantly, machine learning analyses take into consideration the multivariate aspects of
EEG data, and thereby complement traditional univariate analyses that focus on individual
features. Our univariate analyses revealed attentional differences at the sensory level but not
higher cognitive level ERP component. Although the peak N1 amplitude in response to target
tones was greater during on task compared to mind wandering states [33,36], we did not
observe any attentional differences in the N1 and P3 in response to standard tones. In contrast,
Detection of mind wandering using EEG
PLOS ONE | May 12, 2021 13 / 18
our machine learning analyses using ERP components in response to standard tones were suc-
cessful at detecting mind wandering based on these features. These findings highlight the value
of considering multiple features of EEG data in predicting attentional states.
Future work may involve efforts to improve model performance by adding additional data
and more features. Effective machine learning models generally rely on large amounts of data.
Although our proof-of-concept study showed that even with a small sample size, electrophysi-
ological markers can predict mind wandering within subjects and across subjects, these results
need to be replicated in future studies with a larger sample size. Notably, a larger sample size
will provide more training data, and thus it can only enhance accuracy further in the across
subjects analyses. Our finding makes a methodological contribution, namely that accurate
machine learning models of mind wandering can be derived from a sample size of 14 partici-
pants, providing the lower threshold in sample size needed to predict mind wandering above
chance across subjects.
Increasing the number of data points is one way of improving prediction accuracy; another
way is to include multi-modal sources of data. Combining EEG measures with behavioral and
pupillometry features could result in models with better performance, since both types of mea-
sures have resulted in moderately successful models in previous studies [19,21,25]. Other
promising avenues that could improve prediction rates include feature crossing and ensemble
learning, which are techniques that have yet to be implemented often in prior research nor this
current study. The former entails creating synthetic data by multiplying or crossing two or
more features, which may provide predictive abilities beyond those features individually. The
latter is a technique that combines several different machine learning models to minimize
causes of error and improve performance. Another future direction could involve identifying
EEG markers of the types of thoughts we engage in during mind wandering, such as autobio-
graphical memory retrieval or future planning. Results in this present study demonstrate that
SVM and logistic regression classifiers can detect mind wandering; therefore, these models can
potentially be used to further classify task-unrelated thought as a function of its temporal
In summary, this study provides evidence that electrophysiological markers can be
employed in machine learning models to detect mind wandering. Our SVM and logistic
regression classifiers were capable of generalizing across individuals, which has not yet been
demonstrated in other studies that utilize EEG markers. Moreover, our models performed at
above chance levels as determined by several metrics, which is especially promising given that
univariate analyses of the same features did not always show attentional differences. Taken
together, this research brings us closer to the possibility of more intelligent programs that
could detect mind wandering in real-life situations, in real-time.
We thank our participants for their time. We also thank Ludovic Bellier for insightful discus-
sions regarding analyses.
Author Contributions
Conceptualization: Henry W. Dong, Julia W. Y. Kam.
Formal analysis: Henry W. Dong.
Funding acquisition: Robert T. Knight, Julia W. Y. Kam.
Methodology: Henry W. Dong, Caitlin Mills, Julia W. Y. Kam.
Detection of mind wandering using EEG
PLOS ONE | May 12, 2021 14 / 18
Project administration: Henry W. Dong, Julia W. Y. Kam.
Resources: Robert T. Knight.
Supervision: Caitlin Mills, Robert T. Knight, Julia W. Y. Kam.
Validation: Caitlin Mills, Julia W. Y. Kam.
Visualization: Henry W. Dong, Julia W. Y. Kam.
Writing – original draft: Henry W. Dong.
Writing – review & editing: Henry W. Dong, Caitlin Mills, Robert T. Knight, Julia W. Y.
1. Mason MF, Norton MI, Horn JDV, Wegner DM, Grafton ST, Macrae CN. Wandering Minds: The Default
Network and Stimulus-Independent Thought. Science. 2007 Jan 19; 315(5810):393–5.
10.1126/science.1131295 PMID: 17234951
2. Smallwood J, Schooler JW. The restless mind. Psychological Bulletin. 2006; 132(6):946–58. https://doi.
org/10.1037/0033-2909.132.6.946 PMID: 17073528
3. Stawarczyk D, Majerus S, Maj M, Van der Linden M, D’Argembeau A. Mind-wandering: Phenomenol-
ogy and function as assessed with a novel experience sampling method. Acta Psychologica. 2011; 136
(3):370–81. PMID: 21349473
4. Christoff K, Irving ZC, Fox KCR, Spreng RN, Andrews-Hanna JR. Mind-wandering as spontaneous
thought: A dynamic framework. Nature Reviews Neuroscience. 2016; 17(11):718–31.
10.1038/nrn.2016.113 PMID: 27654862
5. Seli P, Kane MJ, Smallwood J, Schacter DL, Maillet D, Schooler JW, et al. Mind-Wandering as a Natural
Kind: A Family-Resemblances View. Trends in Cognitive Sciences. 2018; 22(6):479–90.
10.1016/j.tics.2018.03.010 PMID: 29776466
6. Baird B, Smallwood J, Mrazek MD, Kam JWY, Franklin MS, Schooler JW. Inspired by Distraction: Mind
Wandering Facilitates Creative Incubation. Psychol Sci. 2012 Oct 1; 23(10):1117–22.
1177/0956797612446024 PMID: 22941876
7. Smallwood J, Schooler JW. The Science of Mind Wandering: Empirically Navigating the Stream of Con-
sciousness. Annual Review of Psychology. 2015; 66:487–518.
010814-015331 PMID: 25293689
8. Tan T, Zou H, Chen C, Luo J. Mind Wandering and the Incubation Effect in Insight Problem Solving.
Creativity Research Journal. 2015; 27(4):375–82.
9. Killingsworth MA, Gilbert DT. A wandering mind Is an unhappy mind. Science. 2010;(330):932. https:// PMID: 21071660
10. Song X, Wang X. Mind wandering in Chinese daily lives—an experience sampling study. PLoS ONE.
2012; 7(9):e44423. PMID: 22957071
11. Allan Cheyne J, Solman GJF, Carriere JS a, Smilek D. Anatomy of an error: a bidirectional state model
of task engagement/disengagement and attention-related errors. Cognition. 2009 Apr; 111(1):98–113. PMID: 19215913
12. Kam JWY, Handy TC. The neurocognitive consequences of the wandering mind: a mechanistic account
of sensory-motor decoupling. Frontiers in psychology. 2013 Jan; 4(October):725.
3389/fpsyg.2013.00725 PMID: 24133472
13. Mills C, Graesser A, Risko EF, D’Mello SK. Cognitive coupling during reading. Journal of Experimental
Psychology: General. 2017; 146(6):872–83. PMID: 28447842
14. Schad DJ, Nuthmann A, Engbert R. Your mind wanders weakly, your mind wanders deeply: Objective
measures reveal mindless reading at different levels. Cognition. 2012 Nov 1; 125(2):179–94. https://doi.
org/10.1016/j.cognition.2012.07.004 PMID: 22857818
15. Smallwood J, Fishman DJ, Schooler JW. Counting the cost of an absent mind: Mind wandering as an
underrecognized influence on educational performance. Psychonomic Bulletin and Review. 2007; 14
(2):230–6. PMID: 17694906
16. Weinstein Y. Mind-wandering, how do I measure thee with probes? Let me count the ways. Behav Res
Methods. 2018; 50(2):642–61. PMID: 28643155
Detection of mind wandering using EEG
PLOS ONE | May 12, 2021 15 / 18
17. Vinski MT, Watter S. Priming honesty reduces subjective bias in self-report measures of mind wander-
ing. Conscious Cogn. 2012 Mar; 21(1):451–5. PMID:
18. Mills C, Gregg J, Bixler R, D’Mello SK. Eye-mind reader: An intelligent reading interface that promotes
long-term comprehension by detecting and responding to mind wandering. Human-Computer Interac-
tion. 2020; PMID: 33767571
19. Faber M, Bixler R, D’Mello SK. An automated behavioral measure of mind wandering during computer-
ized reading. Behav Res. 2018 Feb 1; 50(1):134–50. PMID:
20. Hutt S, Mills C, Bosch N, Krasich K, Brockmole J, D’Mello S. “Out of the Fr-Eye-ing Pan”: Towards
Gaze-Based Models of Attention during Learning with Technology in the Classroom. In: Proceedings of
the 25th Conference on User Modeling, Adaptation and Personalization [Internet]. Bratislava, Slovakia:
Association for Computing Machinery; 2017 [cited 2020 Mar 28]. p. 94–103. (UMAP ‘17). Available
21. Mills C, Bixler R, Wang X, D’Mello S. Automatic Gaze-Based Detection of Mind Wandering during Nar-
rative Film Comprehension. International Conference on Educational Data Mining. 2015;30–7.
22. Smilek D, Carriere JSA, Cheyne JA. Out of mind, out of sight: eye blinking as indicator and embodiment
of mind wandering. Psychol Sci. 2010 Jun; 21(6):786–9.
PMID: 20554601
23. Baldwin CL, Roberts DM, Barragan D, Lee JD, Lerner N, Higgins JS. Detecting and Quantifying Mind
Wandering during Simulated Driving. Front Hum Neurosci [Internet]. 2017 [cited 2020 Mar 28]; 11.
Available from: PMID: 28848414
24. Zhang Y, Kumada T. Automatic detection of mind wandering in a simulated driving task with behavioral
measures. PLOS ONE. 2018 Nov 12; 13(11):e0207092.
PMID: 30419060
25. Mills C, D’Mello S. Toward a Real-time (Day) Dreamcatcher: Sensor-Free Detection of Mind Wandering
During Online Reading. International Conference on Educational Data Mining. 2015;
26. Smallwood J, Davies JB, Heim D, Finnigan F, Sudberry M, O’Connor R, et al. Subjective experience
and the attentional lapse: task engagement and disengagement during sustained attention. Conscious-
ness and cognition. 2004 Dec; 13(4):657–90. PMID:
27. Zheng Y, Wang D, Zhang Y, Xu W. Detecting Mind Wandering: An Objective Method via Simultaneous
Control of Respiration and Fingertip Pressure. Front Psychol [Internet]. 2019 [cited 2021 Feb 6]; 10.
Available from: PMID: 30804854
28. Braboszcz C, Delorme A. Lost in thoughts: Neural markers of low alertness during mind wandering.
NeuroImage. 2011 Feb 14; 54(4):3040–7. PMID:
29. Girn M, Mills C, Laycock E, Ellamil M, Ward L, Christoff K. Neural Dynamics of Spontaneous Thought:
An Electroencephalographic Study. In: Proceedings of the 11th International Conference on Aug-
mented Cognition, Lecture Notes in Computer Science. 2017. p. 28–44.
30. Gonc¸alves O
´F, Rêgo G, Conde T, Leite J, Carvalho S, Lapenta OM, et al. Mind Wandering and Task-
Focused Attention: ERP Correlates. Sci Rep [Internet]. 2018 May 15 [cited 2020 Mar 28]; 8. Available
26028-w PMID: 29765144
31. Gruberger M, Simon EB, Levkovitz Y, Zangen A, Hendler T. Towards a Neuroscience of Mind-Wander-
ing. Front Hum Neurosci. 2011; 5(56):1–11. PMID:
32. Hawkins GE, Mittner M, Boekel W, Heathcote A, Forstmann BU. Toward a model-based cognitive neu-
roscience of mind wandering. Neuroscience. 2015 Dec 3; 310:290–305.
neuroscience.2015.09.053 PMID: 26427961
33. Kam JWY, Dao E, Farley J, Fitzpatrick K, Smallwood J, Schooler JW, et al. Slow fluctuations in atten-
tional control of sensory cortex. Journal of cognitive neuroscience. 2011 Feb; 23(2):460–70. https://doi.
org/10.1162/jocn.2010.21443 PMID: 20146593
34. Smallwood J, Beach E, Schooler JW, Handy TC. Going AWOL in the brain: mind wandering reduces
cortical analysis of external events. Journal of cognitive neuroscience. 2008 Mar; 20(3):458–69. https:// PMID: 18004943
35. Baird B, Smallwood J, Lutz A, Schooler JW. The Decoupled Mind: Mind-wandering Disrupts Cortical
Phase-locking to Perceptual Events. Journal of Cognitive Neuroscience. 2014 Apr 17; 26(11):2596–
607. PMID: 24742189
Detection of mind wandering using EEG
PLOS ONE | May 12, 2021 16 / 18
36. Kam JWY, Solbakk AK, Funderud I, Endestad T, Meling TR, Knight RT. Orbitofrontal damage reduces
auditory sensory response in humans. Cortex. 2018; 101:309–12.
2017.12.023 PMID: 29455945
37. Kam JWY, Dao E, Blinn P, Krigolson OE, Boyd LA, Handy TC. Mind wandering and motor control: off-
task thinking disrupts the online adjustment of behavior. Frontiers in human neuroscience. 2012; 6
(December):329. PMID: 23248596
38. O’Connell RG, Dockree PM, Robertson IH, Bellgrove M a, Foxe JJ, Kelly SP. Uncovering the neural sig-
nature of lapsing attention: electrophysiological signals predict errors up to 20 s before they occur. The
Journal of neuroscience: the official journal of the Society for Neuroscience. 2009 Jul 1; 29(26):8604–
11. PMID: 19571151
39. van Son D, De Blasio FM, Fogarty JS, Angelidis A, Barry RJ, Putman P. Frontal EEG theta/beta ratio
during mind wandering episodes. Biological Psychology. 2019; 140(October 2018):19–27. https://doi.
org/10.1016/j.biopsycho.2018.11.003 PMID: 30458199
40. Cavanagh JF, Cohen MX, Allen JJB. Prelude to and Resolution of an Error: EEG Phase Synchrony
Reveals Cognitive Control Dynamics during Action Monitoring. J Neurosci. 2009 Jan 7; 29(1):98–105. PMID: 19129388
41. Cavanagh JF, Zambrano-Vazquez L, Allen JJB. Theta Lingua Franca: A Common Mid-Frontal Sub-
strate for Action Monitoring Processes. Psychophysiology. 2012 Feb; 49(2):220–38.
1111/j.1469-8986.2011.01293.x PMID: 22091878
42. Kam JWY, Solbakk AK, Endestad T, Meling TR, Knight RT. Lateral prefrontal cortex lesion impairs reg-
ulation of internally and externally directed attention. NeuroImage. 2018; 175(February):91–9. https:// PMID: 29604457
43. Kawashima I, Kumano H. Prediction of Mind-Wandering with Electroencephalogram and Non-linear
Regression Modeling. Front Hum Neurosci [Internet]. 2017 Jul 12 [cited 2020 Mar 28]; 11. Available
00365 PMID: 28747879
44. Jin CY, Borst JP, van Vugt MK. Predicting task-general mind-wandering with EEG. Cognitive, Affective
and Behavioral Neuroscience. 2019; PMID: 30850931
45. Jin CY, Borst JP, Vugt MK van. Distinguishing vigilance decrement and low task demands from mind-
wandering: A machine learning analysis of EEG. European Journal of Neuroscience. 2020; 52
(9):4147–64. PMID: 32538509
46. Dhindsa K, Acai A, Wagner N, Bosynak D, Kelly S, Bhandari M, et al. Individualized pattern recognition
for detecting mind wandering from EEG during live lectures. PLOS ONE. 2019 Sep 12; 14(9):
e0222276. PMID: 31513622
47. Tasika NJ, Haque MH, Rimo MB, Haque MA, Alam S, Tamanna T, et al. A Framework for Mind Wander-
ing Detection using EEG Signals. In: 2020 IEEE Region 10 Symposium (TENSYMP). 2020. p. 1474–7.
48. Kam JWY, Dao E, Stanciulescu M, Tildesley H, Handy TC. Mind wandering and the adaptive control of
attentional resources. J Cogn Neurosci. 2013 Jun; 25(6):952–60.
PMID: 23448525
49. Winkler I, Debener S, Mu¨ller K-R, Tangermann M. On the influence of high-pass filtering on ICA-based
artifact reduction in EEG-ERP. Annu Int Conf IEEE Eng Med Biol Soc. 2015; 2015:4101–5. https://doi.
org/10.1109/EMBC.2015.7319296 PMID: 26737196
50. Perrin F, Pernier J, Bertrand O, Echallier JF. Spherical splines for scalp potential and current density
mapping. Electroencephalography and Clinical Neurophysiology. 1989 Feb 1; 72(2):184–7. https://doi.
org/10.1016/0013-4694(89)90180-6 PMID: 2464490
51. Kam JWY, Solbakk A-K, Funderud I, Endestad T, Meling TR, Knight RT. Orbitofrontal damage reduces
auditory sensory response in humans. Cortex. 2018 Apr; 101:309–12.
2017.12.023 PMID: 29455945
52. Delorme A, Makeig S. EEGLAB: an open source toolbox for analysis of single-trial EEG dynamics
including independent component analysis. Journal of neuroscience methods. 2004 Mar 15; 134(1):9–
21. PMID: 15102499
53. Oostenveld R, Fries P, Maris E, Schoffelen JM. FieldTrip: Open source software for advanced analysis
of MEG, EEG, and invasive electrophysiological data. Computational Intelligence and Neuroscience.
2011; 2011:156869. PMID: 21253357
54. Aminuddin MMM, Nasir HM. Focus Loss While Driving Detection by Using Prior Stage ERP as Base-
line. International Journal of Human and Technology Interaction (IJHaTI). 2019 Apr 27; 3(1):39–46–46.
55. Xie S, Wu Y, Zhang Y, Zhang J, Liu C. Single channel single trial P300 detection using extreme learning
machine: Compared with BPNN and SVM. In: 2014 International Joint Conference on Neural Networks
(IJCNN). 2014. p. 544–9.
Detection of mind wandering using EEG
PLOS ONE | May 12, 2021 17 / 18
56. Bougrain L, Saavedra C, Ranta R. Finally, what is the best filter for P300 detection? TOBI Workshop lll-
Tools for Brain-Computer Interaction. 2012;
57. Arevalillo-Herra
´ez M, Cobos M, Roger S, Garcı
´a-Pineda M. Combining Inter-Subject Modeling with a
Subject-Based Data Transformation to Improve Affect Recognition from EEG Signals. Sensors (Basel)
[Internet]. 2019 Jul 8 [cited 2021 Feb 12]; 19(13). Available from:
articles/PMC6651152/ PMID: 31288378
58. Bixler R, D’Mello S. Automatic gaze-based user-independent detection of mind wandering during com-
puterized reading. User Model User-Adap Inter. 2016 Mar 1; 26(1):33–68.
59. Hutt S, Mills C, White S, Donnelly PJ, D’Mello SK. The Eyes Have It: Gaze-Based Detection of Mind
Wandering during Learning with an Intelligent Tutoring System [Internet]. International Educational
Data Mining Society. International Educational Data Mining Society; 2016 [cited 2020 Mar 28]. Available
60. Chawla NiteshV Bowyer KevinW, Hall LawrenceO Kegelmeyer WPhilip. SMOTE: Synthetic Minority
Over-sampling Technique. Journal of Artificial Intelligence Research. 2002 Jun 1; 16:321–57.
61. Blanchard N, Bixler R, Joyce T, D’Mello S. Automated Physiological-Based Detection of Mind Wander-
ing during Learning. In: Trausan-Matu S, Boyer KE, Crosby M, Panourgia K, editors. Intelligent Tutoring
Systems. Cham: Springer International Publishing; 2014. p. 55–60. (Lecture Notes in Computer
62. Schooler JW, Smallwood J, Christoff K, Handy TC, Reichle ED, Sayette MA. Meta-awareness, percep-
tual decoupling and the wandering mind. Trends in Cognitive Sciences. 2011; 15(7):319–26. https://doi.
org/10.1016/j.tics.2011.05.006 PMID: 21684189
Detection of mind wandering using EEG
PLOS ONE | May 12, 2021 18 / 18
... Notably, we do not know of the published studies that searched for the relationship between MW understood as a trait and the organization of functional networks reconstructed on the basis of source-space resting-state EEG. Previous EEG studies aimed, inter alia, at establishing links between MW and the theta/beta ratio 33 , or generally a given frequency peak 34 , and identifying EEG signature of off-task though using thought-probe method and ERP analysis 35,36 . The frequency-focused investigations have produced quite divergent results suggesting that MW is associated with alpha 36,37 delta and gamma 38 or by delta, theta and alpha bands 39 . ...
... In this aspect, functional connections which differentiated our groups have occurred mainly in the theta, delta and gamma bands. The difference regarding frequencies may, on the one hand, result from the different methodological solutions used in our study compared to the previous ones, which often applied a though-probe methodology and task-related EEG recordings analyses, such as ERPs 35 . In addition, other authors ...
... , the source signals from 40 ROIs were evaluated for each epoch. The source signals of each ROI were decomposed into the following five frequency bands: delta (1-4 Hz), theta (4-8 Hz), alpha (8-12 Hz), beta (12-30 Hz), and gamma(30)(31)(32)(33)(34)(35)(36)(37)(38)(39)(40)(41)(42)(43)(44)(45). This signal decomposition was accomplished by using a 6th order zero-phase Butterworth infinite impulse response (IIR) ...
Full-text available
When performing cognitively demanding tasks, people tend to experience momentary distractions or personal associations that intercept their stream of consciousness. This phenomenon is known as Mind Wandering (MW) and it has become a subject of neuroscientific investigations. Off-task thoughts can be analyzed during task performance, but currently, MW is also understood as a dimension of individual differences in cognitive processing. We wanted to recognize the intrinsically-organized functional networks that could be considered the neuronal basis for MW dispositional variability. To achieve this goal we recruited a group of normal adults, and eventually divided the group in half, based on participants’ scores on the scale measuring dispositional MW. Next, these groups were compared regarding the arrangement of preselected intrinsic functional networks, which were reconstructed based on multi-channel signal-source resting-state EEG. It appeared that subjects who tend to mind wander often exhibited decreased synchronization within the default mode network, and, simultaneously, strengthened connectivity between ‘on-task’ networks of diverse functional specificity. Such within- and between networks integrity patterns might suggest that greater Mind Wanderers present an atypical organization of resting-state brain activity, which may translate into attenuated resources needed to maintain attentional control in task-related conditions.
... Previous works have distinguished between mind-wandering and attentive states and achieved a per subject mean accuracy of 65% using SVM and logistic regression and a mean AUC score of 0.715 using SVM and 0.635 using logistic regression. On the leave-one-out participant comparison, they achieved a mean accuracy of 59% using SVM and 58% using logistic regression [3]. ...
... Mind-wandering is often characterized as our attention being oriented away from the task at hand towards our internal, self-generated thoughts. This phenomenon is most often linked to a disruption in normal cognitive functions [3]. Too frequent mind-wandering can lead to depression, anxiety, insomnia, negative mood, and other detrimental effects. ...
In the modern world, it is easy to get lost in thought, partly because of the vast knowledge available at our fingertips via smartphones that divide our cognitive resources and partly because of our intrinsic thoughts. In this work, we aim to find the differences in the neural signatures of mind-wandering and meditation that are common across different meditative styles. We use EEG recording done during meditation sessions by experts of different meditative styles, namely shamatha, zazen, dzogchen, and visualization. We evaluate the models using the leave-one-out validation technique to train on three meditative styles and test the fourth left-out style. With this method, we achieve an average classification accuracy of above 70%, suggesting that EEG signals of meditation techniques have a unique neural signature across meditative styles and can be differentiated from mind-wandering states. In addition, we generate lower-dimensional embeddings from higher-dimensional ones using t-SNE, PCA, and LLE algorithms and observe visual differences in embeddings between meditation and mind-wandering. We also discuss the general flow of the proposed design and contributions to the field of neuro-feedback-enabled mind-wandering detection and correction devices.KeywordsMeditationMind-wanderingClassificationMachine learningDeep learningCognitionNeuro-feedbackEEG
... Recent research demonstrates that it is possible to predict the occurrence of mind wandering from these neurocorrelates. Using logistic regression and support vector machine (SVM), Dong, Mills, Knight, and Kam (2021) found the P3 ERP to be able to predict episodes of mind wandering above chance levels during an auditory tone classification task both within and across participants. Using a support vector machine, Groot et al. (2021) found that widespread increases in delta, theta, and alpha, as well as reduced amplitudes of late Event Related Potentials were most predictive of mind wandering during a SART. ...
... More specifically, we observed a significant attenuation in P3 amplitude, indicating that target stimuli were processed to a lesser extent, but were not completely masked from conscious awareness during mind wandering. Expanding upon previous studies (Baldwin et al., 2017;Barron et al., 2011;Dong et al., 2021;Kam et al., 2011;Smallwood et al., 2008), we observed a relation between mind wandering and reduced conscious processing of rare stimuli during the visuomotor tracking task. These findings demonstrate that cognitive processing of external events is attenuated not only in sustained attention tasks which measure lapses in attention to discrete rare targets, but also during tasks which require continuous motor responses. ...
Full-text available
Fluctuating between external conscious processing and mind wandering is inherent to the human condition. Past research showed that in tasks requiring sustained attention, mind wandering episodes in which attention is directed internally constrain conscious processing of external stimuli. Conversely, conscious processing of internal stimuli is enhanced during mind wandering. To investigate this, we developed and administered a visuomotor tracking task in which participants were instructed to track the path of a stimulus on a screen with a mouse while responding to rare targets. Prior to reports of mind wandering we found the following: The P3 event-related potential component for targets, indicative of conscious stimulus processing, was attenuated at electrodes Cz and Pz. Moreover, alpha power, indicative of internal mental states, increased globally. Theta power increased along the centroparietal area, and beta decreased along right frontal and right centroparietal areas. Interestingly, trait mind wandering was positively correlated with delta power and gamma power, but negatively correlated with the theta-beta ratio. These results demonstrate that mind wandering is characterized by distinct neural signatures at both a state and trait level.
... A solution for this would be the collection of cognitive distraction data in the form of electroencephalographic (EEG) methods to use as a ground truth base to check the TDGV data against. Such systems are able to directly infer and differentiate mind-wandering, external cognitive distraction, and the level of driving task attendance in the driver [24,25]. EEG systems can only be used in experimental settings due to their cumbersome deployment, which heavily restricts their use in real detection systems. ...
Full-text available
A difficult challenge for today’s driver monitoring systems is the detection of cognitive distraction. The present research presents the development of a theory-driven approach for cognitive distraction detection during manual driving based on temporal control theories. It is based solely on changes in the temporal variance of driving-relevant gaze behavior, such as gazes onto the dashboard (TDGV). Validation of the detection method happened in a field and in a simulator study by letting participants drive, alternating with and without a secondary task inducing external cognitive distraction (auditory continuous performance task). The general accuracy of the distraction detection method varies between 68% and 81% based on the quality of an individual prerecorded baseline measurement. As a theory-driven system, it represents not only a step towards a sophisticated cognitive distraction detection method, but also explains that changes in temporal dashboard gaze variance (TDGV) are a useful behavioral indicator for detecting cognitive distraction.
... Given that the denoting directions in space-related spoken speech can evoke many additional associations and distraction from the implemented goal-directed task, this increase in EEG coherence may to some extent be related to the so-called phenomenon of «mind wandering» (MW) [42,17] observed in the process of spontaneous mental activity of a person (Task-Unrelated Thought, TUT). Such states of consciousness are often observed in subjects at rest and may be the basis of stimulus-independent thinking, including creative thinking [4,28]. ...
Although a significant number of studies have been devoted to the investigation of the electrographic correlates and neurophysiological mechanisms of spoken and inner (imagined) speech, there is a question on which EEG characteristics reflect its content. Considering that speech is a complex cognitive process which requires coordinated activity of a number of cortical structures of the large hemispheres, the EEG coherence values were studied. The values were recorded from 14 channels of 10 young men in the process of real verbalization (spoken speech) and during pronunciation of imagined words designating directions in space (up, down, right, left, forward, backward). It was shown that the level of EEG coherence is generally higher for real verbalization, most significantly at gamma-2-rhythm frequencies (55–70 Hz). Spatial coherence patterns specific to a number of words are formed in the left cerebral hemisphere during imagined utterance of words at gamma-2 frequencies. The application of machine learning and neural network classification has demonstrated a significant similarity of the generated spatial coherent patterns of spoken and inner (imagined) speech. The Multi-layer Perceptron (MLP) neural network classification method has shown the accuracy of word detection in the imagined speech according to brain activity patterns up to 49–61% for 3 classes and 33–40% for 7 classes, with a random recognition rate of 33,3 and 14,2% respectively. The latter indicates a promising application of coherence values and imagined speech denoting directions in space for the development of Brain-computer interfaces (BCIs).
Recent advances in computer vision have opened the door for scalable eye tracking using only a webcam. Such solutions are particularly useful for online educational technologies, in which a goal is to respond adaptively to students' ongoing experiences. We used WebGazer, a webcam-based eye-tracker, to automatically detect covert cognitive states during an online reading-comprehension task related to task-unrelated thought and comprehension. We present data from two studies using different populations: (1) a relatively homogenous sample of university students (N = 105), and (2) a more diverse sample from Prolific (N = 173, with < 20% White participants). Across both studies, the webcam-based eye-tracker provided sufficiently accurate and precise gaze measurements to predict both task-unrelated thought and reading comprehension from a single calibration. We also present initial evidence of predictive validity, including a positive correlation between predicted rates of task-unrelated thought and comprehension scores. Finally, we present slicing analyses to determine how performance changed under certain conditions (lighting, glasses, etc.) and generalizability of the results across the two datasets (e.g., training on the data Study 1 and testing on data from Study 2, and vice versa). We conclude by discussing results in the context of remote research and learning technologies.
The ability to mentally wander away from the external environment is a remarkable feature of the human mind. Although recent years have witnessed a surge of interest in examining mind wandering using EEG, there is no comprehensive review that summarizes and accounts for the variable findings. Accordingly, we conducted a systematic review that synthesizes evidence from EEG studies that examined the electrophysiological measures of mind wandering. Our search yielded 42 studies that met eligibility criteria. The reviewed literature converges on a reduction in the amplitude of canonical ERP components (i.e., P1, N1 and P3) as the most reliable markers of mind wandering. Spectral findings were less robust, but point towards greater activity in lower frequency bands, (i.e., delta, theta, and alpha), as well as a decrease in beta band activity, during mind wandering compared to on-task states. The variability in these findings appears to be modulated by the task context. To integrate these findings, we propose an electrophysiological account of mind wandering that explains how the brain supports this experience. Conclusions drawn from this work will inform future endeavours in basic science to map out electrophysiological patterns underlying mind wandering and in translational science using EEG to predict the occurrence of this phenomenon.
Full-text available
Mind‐wandering is a ubiquitous mental phenomenon that is defined as self‐generated thought irrelevant to the ongoing task. Mind‐wandering tends to occur when people are in a low‐vigilance state or when they are performing a very easy task. In the current study, we investigated whether mind‐wandering is completely dependent on vigilance and current task demands, or whether it is an independent phenomenon. To this end, we trained support vector machine (SVM) classifiers on EEG data in conditions of low and high vigilance, as well as under conditions of low and high task demands, and subsequently tested those classifiers on participants’ self‐reported mind‐wandering. Participants’ momentary mental state was measured by means of intermittent thought probes in which they reported on their current mental state. The results showed that neither the Vigilance classifier nor the Task demands classifier could predict mind wandering above chance level, while a classifier trained on self‐reports of mind‐wandering was able to do so. This suggests that mind‐wandering is a mental state different from low vigilance or performing tasks with low demands—both which could be discriminated from the EEG above chance. Furthermore, we used dipole fitting to source‐localize the neural correlates of the most import features in each of the three classifiers, indeed finding a few shared neural structures among the three phenomena. Our study demonstrates the value of machine learning classifiers in unveiling patterns in neural data and uncovering the associated neural structures by combining it with an EEG source analysis technique.
Full-text available
Neural correlates of mind wandering: The ability to detect mind wandering as it occurs is an important step towards improving our understanding of this phenomenon and studying its effects on learning and performance. Current detection methods typically rely on observable behaviour in laboratory settings, which do not capture the underlying neural processes and may not translate well into real-world settings. We address both of these issues by recording electroencephalography (EEG) simultaneously from 15 participants during live lectures on research in orthopedic surgery. We performed traditional group-level analysis and found neural correlates of mind wandering during live lectures that are similar to those found in some laboratory studies, including a decrease in occipitoparietal alpha power and frontal, temporal, and occipital beta power. However, individual-level analysis of these same data revealed that patterns of brain activity associated with mind wandering were more broadly distributed and highly individualized than revealed in the group-level analysis. Mind wandering detection: To apply these findings to mind wandering detection, we used a data-driven method known as common spatial patterns to discover scalp topologies for each individual that reflects their differences in brain activity when mind wandering versus attending to lectures. This approach avoids reliance on known neural correlates primarily established through group-level statistics. Using this method for individual-level machine learning of mind wandering from EEG, we were able to achieve an average detection accuracy of 80-83%. Conclusions: Modelling mind wandering at the individual level may reveal important details about its neural correlates that are not reflected when using traditional observational and statistical methods. Using machine learning techniques for this purpose can provide new insight into the varieties of neural activity involved in mind wandering, while also enabling real-time detection of mind wandering in naturalistic settings.
Full-text available
Existing correlations between features extracted from Electroencephalography (EEG) signals and emotional aspects have motivated the development of a diversity of EEG-based affect detection methods. Both intra-subject and inter-subject approaches have been used in this context. Intra-subject approaches generally suffer from the small sample problem, and require the collection of exhaustive data for each new user before the detection system is usable. On the contrary, inter-subject models do not account for the personality and physiological influence of how the individual is feeling and expressing emotions. In this paper, we analyze both modeling approaches, using three public repositories. The results show that the subject’s influence on the EEG signals is substantially higher than that of the emotion and hence it is necessary to account for the subject’s influence on the EEG signals. To do this, we propose a data transformation that seamlessly integrates individual traits into an inter-subject approach, improving classification results.
Full-text available
Mind-wandering refers to the process of thinking task-unrelated thoughts while performing a task. The dynamics of mind-wandering remain elusive because it is difficult to track when someone’s mind is wandering based only on behavior. The goal of this study is to develop a machine-learning classifier that can determine someone’s mind-wandering state online using electroencephalography (EEG) in a way that generalizes across tasks. In particular, we trained machine-learning models on EEG markers to classify the participants’ current state as either mind-wandering or on-task. To be able to examine the task generality of the classifier, two different paradigms were adopted in this study: a sustained attention to response task (SART) and a visual search task. In both tasks, probe questions asking for a self-report of the thoughts at that moment were inserted at random moments, and participants’ responses to the probes were used to create labels for the classifier. The 6 trials preceding an off-task response were labeled as mind-wandering, whereas the 6 trials predicting an on-task response were labeled as on-task. The EEG markers used as features for the classifier included single-trial P1, N1, and P3, the power and coherence in the theta (4–8 Hz) and alpha (8.5–12 Hz) bands at PO7, Pz, PO8, and Fz. We used a support vector machine as the training algorithm to learn the connection between EEG markers and the current mind-wandering state. We were able to distinguish between on-task and off-task thinking with an accuracy ranging from 0.50 to 0.85. Moreover, the classifiers were task-general: The average accuracy in across-task prediction was 60%, which was above chance level. Among all the extracted EEG markers, alpha power was most predictive of mind-wandering.
Full-text available
Mind wandering happens when one train of thought, related to a current undertaking, is interrupted by unrelated thoughts. The detection and evaluation of mind wandering can greatly help in understanding the attention control mechanism during certain focal tasks. Subjective assessments such as random thought-probe and spontaneous self-report are the ways previous research has assessed mind wandering. Here we propose a task in which participants are asked to simultaneously control respiration and fingertip pressure. They are instructed to click a force sensor at the exact moment of inhalation and exhalation of their respiration. The temporal synchronization between the respiratory signals and the fingertip force pulses offers an objective index to detect mind wandering. Twelve participants engaged in the proposed task in which self-reports of mind wandering are compared with the proposed objective index. The results show that the participants reported significantly more mind-wandering episodes during the trials with a larger temporal synchronization than they did during those trials with a smaller temporal synchronization. The findings suggest that the temporal synchronization might be used as an objective marker of mind wandering in attention training and exploration of the attention control mechanism.
Full-text available
Mind wandering (MW) is extremely common during driving and is often accompanied by performance losses. This study investigated the use of driving behavior measurements to automatically detect mind wandering state in the driving task. In the experiment, participants (N = 40) performed a car-following task in a driving simulator and reported, upon hearing a tone, whether they were experiencing mind wandering or not. Supervised machine learning techniques were applied to classify MW-absent versus MW-present state, using both driver-independent and driver-dependent modeling methods. In the driver-independent modeling, we separately built models for participants with high or low MW and participants with medium MW. The optimal models can not offer a significant improvement than other models. So building effective driver-independent models with the leave-one-participant-out cross-validation method is challenging. In the driver-dependent modeling, we built models for each participant with medium MW. The best models of some participants were effective. The results indicate the development of mind wandering detecting system should take into account both inter-individual and intra-individual difference. This study provides a step toward minimizing the negative impacts of mindless driving and should benefit other fields of psychological research.
Conference Paper
Mind Wandering (MW) is the recurrent occurrence in which our mind gets disengaged from the immediate task and focused on internal trains of thought. MW can have both good as well as detrimental effects. Hence, it is crucial to measure MW. This interesting phenomenon and part of our daily life can be effectively measured using EEG signals. Several techniques that have been used to predict MW. However, literature shows that there are still chances of further improvement in this field. Therefore, in this paper we proposed a framework based on data mining and machine learning to detect MW using EEG signals. In our framework, we extracted a number of features EEG channels. We evaluate the performance of our proposed framework using 19 sessions of two subjects. The accuracy of the proposed framework is higher than the other researches under this field that indicates the superiority of our proposed framework.
We zone out roughly 20-40% of the time during reading – a rate that is concerning given the negative relationship between mind-wandering and comprehension. We tested if Eye-Mind Reader – an intelligent interface that targeted mind-wandering as it occurred – could mitigate its negative impact on reading comprehension. When an eye-gaze-based classifier indicated that a reader was mind-wandering, those in a MW-Intervention condition were asked to self-explain the concept they were reading about. If the self-explanation quality was deemed subpar by an automated scoring mechanism, readers were asked to re-read parts of the text in order to correct their comprehension deficits and improve their self-explanation. Each participant in the MW-Intervention condition was paired with a Yoked-Control counterpart who received the exact same interventions regardless of whether they were mind-wandering. Results indicate that re-reading improved self-explanation quality for the MW-Intervention group, but not the control group. The two conditions performed equally well on textbase (i.e. fact-based) and inference-level comprehension questions immediately after reading. However, after a week-long delay, the MW-Intervention condition significantly outperformed the yoked-control condition on both comprehension assessments (ds = .352 and .307). Our findings suggest that real-time interventions during critical periods of mind-wandering can promote long-term retention and comprehension.
Background: In resting-state EEG, the ratio between frontal power in the slow theta frequency band and the fast beta frequency band (the theta/beta ratio, TBR) has previously been negatively related to attentional control. Also, increased theta and reduced beta power were observed during mind wandering (MW) compared to episodes of focused attention. Thus, increased resting-state frontal TBR could be related to MW, suggesting that previously observed relationships between TBR and attentional control could reflect MW episodes increasing the average resting state TBR in people with low attentional control. Goals: To replicate and extend the previous theta and beta MW effects for frontal TBR recordings and test if MW related changes in frontal TBR are related to attentional control. Method: Twenty-six healthy participants performed a 40-minute breath-counting task, after a baseline EEG recording, while EEG was measured and participants indicated MW episodes with button presses. Results: Frontal TBR was significantly higher during MW episodes than during on-task periods. However, no relation between frontal TBR and attentional control was found. Conclusions: This confirms that frontal TBR varies with MW, which is thought to reflect, among other things, a state of reduced top-down attentional control over thoughts. 194 words.
As empirical research on mind-wandering accelerates, we draw attention to an emerging trend in how mind-wandering is conceptualized. Previously articulated definitions of mind-wandering differ from each other in important ways, yet they also maintain overlapping characteristics. This conceptual structure suggests that mind-wandering is best considered from a family-resemblances perspective, which entails treating it as a graded, heterogeneous construct and clearly measuring and describing the specific aspect(s) of mind-wandering that researchers are investigating. We believe that adopting this family-resemblances approach will increase conceptual and methodological connections among related phenomena in the mind-wandering family and encourage a more nuanced and precise understanding of the many varieties of mind-wandering.