Brain Stimulation 8 (2015) 253e259

Contents lists available at ScienceDirect

Brain Stimulation
journal homepage: www.brainstimjrnl.com

Original Articles

Its the Thought That Counts: Examining the Task-dependent

Effects of Transcranial Direct Current Stimulation on Executive
Function
Jay Gill a, b, Priyanka P. Shah-Basak a, Roy Hamilton a, b, c, *
a

Laboratory for Cognition and Neural Stimulation, University of Pennsylvania, Philadelphia, PA, USA
Center for Cognitive Neuroscience, University of Pennsylvania, Philadelphia, PA, USA
c
Department of Neurology, University of Pennsylvania, Philadelphia, PA, USA
b

a r t i c l e i n f o

a b s t r a c t

Article history:
Received 26 August 2014
Accepted 29 October 2014
Available online 25 November 2014

Background: Prior investigations employing transcranial direct current stimulation (tDCS) have shown
that stimulation can elicit subsequent improvement on tests of various cognitive abilities, including
working memory. While stimulation parameters such as intensity and duration are known to determine
the effects of tDCS, the degree to which stimulation effects are inuenced by the nature of cognitive
activities during stimulation remains unclear.
Objective/hypothesis: To determine whether manipulating the working memory load of a task performed
during stimulation would modulate tDCS-induced enhancement of performance on a different, related
measure after stimulation.
Methods: In two separate but closely related sham-controlled experiments, two groups of healthy
subjects underwent anodal tDCS (2 mA) of the left dorsolateral prefrontal cortex (DLPFC) for 20 min. In
Experiment 1, subjects (n 11) trained on a letter 3Back task during stimulation. In Experiment 2
subjects (n 11) trained on a letter 1Back task, which resembled the 3Back task but featured a lower
working memory load. In both experiments, before and after stimulation, subjects completed an
adjusting Paced Auditory Serial Addition Task (A-PASAT). Both the experimenter and subjects were blind
to stimulation conditions in both experiments.
Results: Subjects were both faster and more accurate on the A-PASAT task after receiving real tDCS paired
with 3Back training (Experiment1) compared to sham3Back, real1Back, and sham1Back conditions.
Conclusions: The cognitive demands of a task performed during tDCS can inuence the effects of tDCS on
post-stimulation performance. This nding has direct relevance to the use of tDCS as an investigative tool
in cognitive neuroscience and as a therapy.
2015 Elsevier Inc. All rights reserved.

Keywords:
tDCS
Working memory
Cognitive enhancement
Cognitive remediation

Introduction
While a growing body of experiments have shown that transcranial direct current stimulation (tDCS) can be employed to
modulate a variety of complex cognitive abilities [1e4], the
methods employed in this large and growing body of work remain
heterogenous. This variability reects, in part, incomplete understanding of how different properties of tDCS such as current intensity [5], duration of stimulation, electrode placement, and

* Corresponding author. Goddard Laboratories, Room 518, University of Pennsylvania, 3710 Hamilton Walk, Philadelphia, PA 19104, USA. Tel.: 1 215 573 7090;
fax: 1 215 898 1982.
E-mail address: Roy.Hamilton@uphs.upenn.edu (R. Hamilton).
http://dx.doi.org/10.1016/j.brs.2014.10.018
1935-861X/ 2015 Elsevier Inc. All rights reserved.

direction of current ow impact changes in neural excitability [6]

and cognitive function [3,7,8]. One potential determinant of stimulation effects that remains to be fully characterized is taskdependency, that is the degree to which the cognitive or behavioral activity that an individual is engaged in during tDCS inuences
the subsequent neurophysiologic and behavioral effects of stimulation [8e10].
Numerous studies have paired tDCS with specic cognitive or
behavioral tasks [5,11,12], operating under the assumption that a
complementary or additive relationship exists between stimulation
and task rehearsal. On one level the logic of this approach is intuitive: if tDCS and rehearsal independently improve task performance, then it stands to reason that combining them may lead to
further enhanced performance changes. However, a potentially

254

J. Gill et al. / Brain Stimulation 8 (2015) 253e259

more nuanced argument in favor of this approach is that tDCS, by

virtue of its relatively subtle modulatory effects on neural activity,
may be more effective in brain networks that are already selectively
engaged by a cognitive activity. By that account, cognitive and
behavioral tasks used in conjunction with tDCS may play a determinative role in which brain regions, connections, and pathways
are most affected by stimulation [13]. Extending this reasoning
further, one could hypothesize that pairing stimulation with tasks
that engage specic cognitive domains would result in selective
changes in brain activity, the magnitude of which would be
dependent on the degree to which the tasks paired with stimulation engaged the cognitive domains and brain networks in question. However, few studies to date have explored whether tDCS
paired with behavioral tasks results in change in performance on
cognitively related tasks [9,10], and no study has directly tested
whether the effects of tDCS can be inuenced by varying the level of
domain-specic cognitive demand of a concurrent training task.
We sought to examine the role of training task-dependency in
enhancing the effects of tDCS on executive function abilities in
healthy adults. In a two-part experiment (Fig. 1), we explored
whether varying the degree to which working memory was
engaged during stimulation affected the ability of anodal tDCS to
transiently enhance working memory performance after stimulation. Prior work in normal controls has indicated the crucial role of
the left dorsolateral prefrontal cortex (DLPFC) in working memory
functions [14e16] as well as other executive functions [17e19].
Therefore, in Experiment 1, subjects received anodal tDCS to the left
DLPFC [1] while completing a letter 3Back working memory task. In
Experiment 2, a different group of subjects underwent an identical
stimulation paradigm but performed a 1Back taskda task that was
identical in design to the prior task except less dependent on
working memory functiondduring stimulation. Due to the higher
working memory demand of the 3Back compared to the 1Back, we
predicted that anodal tDCS combined with training with the 3Back
task would be associated with transient improvement on a different
task that required working memory. We tested this hypothesis by
administering an adjusting version of the Paced Auditory Serial
Addition Test (A-PASAT) before and after stimulation [20].

Materials and methods

TDCS was delivered by a constant-current battery operated

stimulator unit (Magstim Eldith 1 Channel DC Stimulator Plus,

Counterbalanced
Anodal tDCS
Sham

Visit 1

Visit 2
> 48 hours

A-PASAT 3 Back A-PASAT

A-PASAT 3 Back

Visit 1

Visit 2

Experiment 2

1 Back

We employed an adjusting version of the Paced Auditory

Serial Addition Task (A-PASAT) [20] to assess changes in performance immediately after stimulation in both Experiments 1 and
2. Single digits ranging from 1 to 9 were presented aurally.
Subjects were instructed to sum the two most recently presented
numbers and respond verbally prior to the presentation of the
next digit. Interstimulus intervals (ISIs) were altered depending
on subjects performance. When a subject answered correctly
within the stimulus presentation window prior to the appearance of the next digit, the ISI was reduced by 20 ms; when a
subject answered incorrectly, the ISI increased by 20 ms. The
number stimuli were comprised of pre-recorded audio les
(500 ms in duration) that were played using the E-Prime E Studio
(v1.2 Psychology Software Tools, Inc.) on a laptop computer (Dell
Latitude E6400). The task was composed of 4 blocks of 60 trials.
The initial ISI was 2400 ms at the beginning of block 1, and was
subsequently adjusted based on subject accuracy. Subject responses were captured using a digital audio recorder and scored
ofine. Performance was scored with respect to overall accuracy.
Accuracy was also evaluated as a function of ISI, ranging from
2400 ms up to 220 ms. For the purposes of this analysis, ISIs were
grouped into intervals of 200 ms. Lastly, we evaluated trial frequencies, which is the number of trials that subjects completed
in each ISI grouping.

A-PASAT

A-PASAT 1 Back

Counterbalanced

Figure 1. Overview of Experiments 1 and 2.

Subjects
Eleven right-handed young adults (3 females) ages 18e25 years
(21.8  2.7 years) with no history of neurological or psychiatric
disorders were enrolled in Experiment 1. Subjects completed an
average of 15.6 years of education (SD 1.5). None were taking
prescribed anti-depressants or other psychoactive medications, and
none had contraindications to receiving tDCS. This study was
approved by the Institutional Review Board of the University of
Pennsylvania. All subjects provided informed consent.

A-PASAT

> 48 hours
A-PASAT

Adjusting Paced Auditory Serial Addition Task (A-PASAT)

Experiment 1

Stimulation parameters

Experiment 1

Magstim, Whitland, UK) via two 5  5 cm2 electrodes soaked in a

saline solution. All subjects received both anodal and sham stimulation for 20-min in 2 separate sessions (3.92  3.7 days separated
these sessions). The active electrode was placed over F3 using the
10-20 International EEG system, a region corresponding to the left
DLPFC. The reference electrode was placed over the right supraorbital region. Anodal stimulation was delivered at 2 mA for 20-min
(current density: 0.80 mA/mm2). During sham stimulation, current
was ramped up to 2 mA and then back down to 0 mA in the rst
30 s, which remained at 0 mA for the rest of the 20 min period. The
order in which subjects received real and sham stimulation was
counterbalanced. Both subjects and the experimenter were blinded
to the type of stimulation applied during each session.

A-PASAT

Training task e letter 3Back

Subjects completed the letter 3Back task during stimulation in
Experiment 1 (Fig. 1). Stimuli were displayed on a laptop computer
(Dell Latitude E6400) centered in front of the subject and were
generated using E-prime E Studio (v1.2 Psychology Software Tools,
Inc.). The task consisted of 10 lower-case letters [a, b, c, d, e, i, l, o, p,
q] presented in a pseudorandom order over 764 trials (20-min).
Letters that resembled each other (homolyphs) were intentionally
incorporated and presented in sequence to increase the difculty of
the task. Each letter was displayed on the screen for 1300 ms and
was followed by a 50 ms blank screen. Subjects were instructed to
press 2 if the letter on the screen matched with the letter that

J. Gill et al. / Brain Stimulation 8 (2015) 253e259

nBack

Time

a
o

1 Back target
3 Back target

a
Figure 2. Training task: Letter nBack performed during stimulation; 3Back was carried
out in Experiment 1 and 1Back was carried out in Experiment 2.

occurred 3 trials prior (Fig. 2). If the letter did not match, subjects
were instructed to press 1 while the letter was still on the screen.
A response was required for each stimulus that appeared on the
screen. For the 20-min stimulation period, the task consisted of 43%
targets (328 trials) and was divided into four 191 trial (4.25 min)
blocks, separated by 1-min breaks. During breaks, current was still
delivered and subjects cumulative accuracy up to that point was
displayed on the screen. To encourage adequate subject effort, those
who achieved a 75% or higher cumulative accuracy received additional compensation ($10). To reduce the inuence of practice
effects, 2 different versions of the 3Back were generated. The order
in which subjects performed different versions of the 3back was
counterbalanced.
Experiment 2

255

(in Experiment 1) or the 1Back (in Experiment 2) during stimulation, subjects practiced the task for 22 trials (7 targets), lasting
about 2-min, to ensure that they understood and followed the instructions appropriately. During practice, an accuracy of 65% was
required for the 3Back and 90% accuracy was required for the 1Back,
before subjects could advance to the stimulation sessions. Similarly,
during subjects rst visits, they practiced on the A-PASAT for 200
trials to familiarize them with the task. However, an accuracy
threshold was not implemented during practice for this task. In
Experiment 1, subjects completed practice trials, followed by a full
version of the A-PASAT. Subjects then received 20-min of stimulation (real or sham) and concurrent training with the 3Back task,
followed by completion of a second full version of the A-PASAT. For
the subsequent visit, subjects completed another full version of the
A-PASAT, followed by the stimulation (sham or real) that they did
not receive during their rst visit, again paired with the 3Back task,
and followed by another administration of the A-PASAT. Experiment 2 shared the same sequence of events, except that subjects
received training on the 1Back task concurrently with tDCS instead
of the 3Back (Fig. 1).
Results
SPSS v. 20 and R statistical packages were employed to analyze
and display the behavioral data.
To examine the main hypothesis that anodal stimulation paired
with an engaging working memory training task (3Back) would
facilitate subsequent working memory performance more than a
less demanding training task (1Back), we ran a 2  2 mixed design
ANOVA with change in the overall A-PASAT accuracy (across all ISIs)
as the dependent variable. The results of Experiment 1 and 2 were
combined for this analysis. Stimulation condition (real anodal,
sham) was included as the within-subject factor and the training

Subjects
Twelve right-handed young adults (5 female) ages of 18e25
(19.8  1.5 years) with no history of neurological or psychiatric
disorders were enrolled in Experiment 2; one subject had to be
excluded because of technical errors. Included subjects completed
an average of 14.1 years (1.0) years of education. The exclusion
criteria were identical to those used in Experiment 1.
Training task e letter 1Back
Subjects completed the 1Back task during stimulation in
Experiment 2 (Fig. 1). Equipment and software for displaying
stimuli were the same as in Experiment 1. The task consisted of 10
lower-case letters [a, b, j, z, k, t, o, p, x, r] presented in pseudorandom order over 764 trials (20-min). As in Experiment 1, stimuli
ashed on the screen for 1300 ms, followed by a 50 ms blank
screen. Subjects were instructed to identify whether a letter had
occurred twice in a row and press 2 while the second letter was
still on the screen (Fig. 2); otherwise, they were instructed to press
1. To maintain consistency with the 3Back task in Experiment 1,
the task was broken into 4, 4.25 min blocks separated by 3, 1-min
breaks. During breaks, subjects were provided with their cumulative accuracy. In line with 3Back training task, 2 different versions of
the 1Back were generated to reduce practice effects. The order in
which subjects performed different versions of the 1back was
counterbalanced.
Procedures for Experiments 1 and 2
During the rst visit, all subjects completed practice sessions
with the 3Back in Experiment 1, the 1Back in Experiment 2 and the
A-PASAT in both Experiment 1 and 2. Prior to performing the 3Back

Figure 3. Change in overall accuracy of the A-PASATdthe task performed immediately

after stimulation with real anodal or sham conditions in Experiment 1 and 2. 3Back
was performed during stimulation (training task) in Experiment 1, and 1Back was
performed during stimulation in Experiment 2. Vertical lines represent standard errors,
* reects statistical signicance (P < 0.05), and y reects trend toward signicance
(P < 0.1).

256

J. Gill et al. / Brain Stimulation 8 (2015) 253e259

Figure 4. Experiment 1: (A) Change in A-PASAT accuracy (%) per ISI group after training with 3Back task during stimulation. Based on subjects performance at baseline, the ISI
groups consisting of ISI range 2400e1100 ms was characterized as Hard conditions, and those consisting of 1080e220 ms as Easy conditions. (B) Grouping Hard and Easy ISI
conditions into Hard and Easy difculty groups. (C) A-PASAT trial frequencies across all ISI groups. Vertical lines represent standard errors, * reects statistical signicance at
P < 0.05), ** reects statistical signicance at P < 0.01, and y reects trend toward signicance (P < 0.1).

task (3Back, 1Back) was included as the between-subject factor; we

also examined the interaction between these factors. This analysis
revealed that although the main effects of stimulation and task
performed during stimulation were not signicant (P > 0.05), the
stimulation by task interaction was highly signicant (F1,20 8.2,
P 0.01; Fig. 3). Posthoc paired t-tests indicated that the performance on A-PASAT improved signicantly (d 25.6%; t 2.3,
P 0.023) after real stimulation as compared to the sham only
while training with the 3Back, and that overall performance tended
to be better after sham stimulation (d 13.1%; t 1.74, P 0.056)
while training with the 1Back.
Next, to further disambiguate the role of cognitive load of the
training tasks, we explored the effects of anodal stimulation on
A-PASAT performance as a function of ISIs, separately for Experiment 1 and 2.
Experiment 1
The average accuracy on the 3Back task performed while
receiving real anodal tDCS was 76.0% (13.5%), and while receiving

sham tDCS was 74.3 (11.8%); paired t-test revealed that the performance did not differ between real anodal and sham tDCS conditions (P > 0.05).
As expected, baseline accuracy diminished as the difculty of
A-PASAT increased (with the shortening of the ISIs) across the ISI
groups. Specically, baseline accuracy for ISI groups ranging from
1080 to 220 ms was below 75%; we characterized these ISI groups
as Hard conditions (average accuracy: 49.1  32%; range:
71.7e21.8%). ISI groups comprising 2400e1100 ms, for which the
average baseline accuracy was 75% or above, were characterized as
Easy conditions (average accuracy: 91.5  12%; range:
98.7e85.4%). We next evaluated differences in A-PASAT accuracies
across different ISI groups between the real anodal and sham
stimulation conditions. A repeated measures ANOVA with ISI
groups (2400e2200, ., 420e220) and stimulation (real, sham) as
xed factors revealed signicant main effects (P < 0.05) as well as
interaction between these variables (F9,90 2.28, P 0.024).
Posthoc paired t-tests revealed that in 3 out of 4 Hard ISI conditions,
differences in accuracy were greater in the real stimulation condition compared to the sham condition (P < 0.05; Fig. 4A); for a single

J. Gill et al. / Brain Stimulation 8 (2015) 253e259

Hard condition (1080e880 ms), there was a trend toward improved

performance (d 13.5%; t 1.7, P 0.063). The performance across
Easy ISI conditions did not differ between the real stimulation and
the sham stimulation conditions (all Ps > 0.05). To simplify the
interpretation of the stimulation-specic effects, we collapsed the
data across the Hard and Easy ISI conditions in order to form Hard
and Easy difculty groups, respectively (Fig. 4B). Comparison of
performance across stimulation conditions and task difculty
groups indicated that anodal stimulation paired with 3Back task,
improved performance specically on harder trials of the A-PASAT
(d 15.8%; t 4.26, P < 0.001).
Lastly, the increased accuracy on harder A-PASAT trials is also
reected in a shift in trial frequencies following anodal stimulation
(N 9; Fig. 4C). Following anodal tDCS, trials within the Hard ISI
conditions occurred more frequently compared to pre-anodal
stimulation as well as pre- and post-sham conditions. A repeated
measures ANOVA comparing the trial distribution across stimulation conditions (real, sham), session (pre, post) and ISI groups
revealed signicant main effects of stimulation (F1,8 9.71,
P 0.014), session (F1,8 9.71, P 0.014) and ISI groups
(F1,8 97.28, P < 0.001), and also a signicant three-way interaction
among these variables (F1,8 16.4, P 0.004). Posthoc paired t-tests
revealed there was a strong trend toward a signicant difference in
trial frequencies post-versus pre-real stimulation (t 2.11,
P 0.067) for the 1080e880 ms ISI group, rst of the Hard ISI
conditions, however for post-versus pre-sham stimulation conditions, this difference was not signicant (d 1.22, P > 0.05). In
addition, posthoc paired t-tests revealed that there was a signicant
difference in trial frequency in the 420 220 ms ISI group, last of the
Hard ISI conditions, following real anodal tDCS (t 2.589,
P 0.032) compared to the sham condition (P > 0.05). Together,
these ndings suggest that after receiving real stimulation, subjects
tended to require fewer trials to reach to Hard ISI conditions, and
spent many more trials in the most difcult ISI group among the
Hard conditions.
Experiment 2
Subjects were highly accurate on the 1Back task irrespective of
stimulation condition (real anodal: 94.4  5%; sham: 96  3%);

257

paired t-test revealed that the difference between real anodal and
sham tDCS conditions was not signicant (P > 0.05).
Accuracy diminished as the difculty level of A-PASAT increased
(with the shortening of the ISIs) across the ISI groups. The classication of Hard and Easy ISI conditions based on the average
baseline accuracy was similar to Experiment 1. The average accuracy at baseline for Hard ISI conditions was 39.7  30.1% (range:
67.3e5.4%), and that for the Easy conditions was 91.7  11.8%
(range: 96.4e88.9%).
A repeated measures ANOVA with ISI groups (2400e2200, .,
420e220) and stimulation (real, sham) revealed that none of the
main effects were signicant, and also the interaction between ISI
groups and stimulation was not signicant (all P > 0.05; Fig. 5A).
Although this interaction was not signicant, we compared the
performance across stimulation conditions in the Hard difculty
group, which revealed a trend toward somewhat better performance after real stimulation compared to the sham (d 5.8%;
t 1.32, P 0.097; Fig. 5B).
Discussion
We provide compelling evidence that, in addition to other
well-established stimulation parameters, the cognitive enhancement capacity of tDCS may also depend on the nature of the task
that is performed during stimulation. When administered concurrently with anodal tDCS of the left DLPFC, the letter 3Backda
task that robustly engages networks involved in working
memorydincreased prociency on the A-PASAT, a separate task
that relies on working memory. By contrast, when the subjects were
trained on the 1Backda task that is less engaging than the 3Back
with respect to working memorydno improvement on the
A-PASAT was observed. Insofar as the only difference between
the two experimental conditions was the working memory load of
the task performed during stimulation, we argue that the positive
effects of tDCS were dependent on the cognitive demands of the
task administered concurrently with stimulation.
Functional neuroimaging studies demonstrate that activation of
working memory-related brain regions, including the left DLPFC, is
highly correlated with the difculty of the nBack task (as indicated
by the size of n) [16,21]. We suggest that in this study endogenous

Figure 5. Experiment 2: (A) Change in A-PASAT accuracy (%) per ISI group after training with 1Back task during stimulation. Based on subjects performance at baseline, the ISI
groups consisting of ISI range 2400e1100 ms was characterized as Hard conditions, and those consisting of 1080e220 ms as Easy conditions. (B) Grouping Hard and Easy ISI
conditions into Hard and Easy difculty groups. Vertical lines represent standard errors, and y reects trend toward signicance (P < 0.1).

258

J. Gill et al. / Brain Stimulation 8 (2015) 253e259

(3Back) and exogenous (anodal tDCS) modulation of left DLPFC

activity synergistically paired to induce robust post-stimulation
behavioral improvement. By contrast, this synergistic pairing was
absent in Experiment 2, since the low working memory load of the
1Back task may not have engaged working memory networks as
strongly.
The concept of task-dependency is consistent with accepted
notions of how tDCS inuences neural activity. The current
employed in tDCS is insufcient to directly stimulate action
potentials in neurons, and is instead believed to induce incremental
shifts in neuronal membrane potentials that alter the likelihood of
neurons ring over time [22,23]. Extending this idea, these subthreshold shifts in the probability of neuronal ring may preferentially affect networks of neurons that are already engaged in
activity compared to networks at relative rest [13]. By this account,
pairing tDCS with specic tasks may provide an important way to
accentuate both the behavioral effects of stimulation as well as the
anatomic specicity a technology that is otherwise thought to have
low spatial resolution [24].
Our ndings build upon prior investigations that have
combined tDCS with training on cognitive tasks. In many instances the tasks that have been paired with stimulation were
specically those that were shown to improve after treatment
(e.g., Refs. [1,25,26]). However, in a much smaller number of
investigations, the efcacy of pairing tDCS with concurrent
training tasks was explored by examining their effects on
different tasks that engaged similar abilities within a cognitive
domain (e.g., Refs. [27,28]). One study by Andrews and colleagues (2011) is especially germane to our results, in that the
investigators demonstrated that anodal tDCS paired with an
nBack task resulted in improved performance on a different
working memory task (digit span), and also showed that there
was no improvement in performance when tDCS was delivered
in the absence of any behavioral task [9]. By employing
experimental conditions that varied specically in working
memory load, we have extended these prior results by further
clarifying that the effects of pairing tDCS with a behavioral task
are mediated by the domain-specic cognitive demands of the
training task.
It is noteworthy that, while both the 3Back task and A-PASAT
engage working memory, there are differences in the cognitive
demands of the two tasks that may have come into play in this
study. Specically, the A-PASAT relies upon other frontal-executive
functions in addition to working memory, such as selective attention, information processing speed, and set switching [20,29e31].
In this study, we observed that when subjects received real tDCS in
conjunction with the 3Back task, they experienced a pronounced
improvement in their ability to perform the A-PASAT accurately at
very short ISIs. Consistent with prior evidence that suggests that
tDCS paired with working memory training can exhibit near
transfer to closely related fronto-executive tasks [28], it is possible
that this aspect of our subjects improved performance may have
stemmed from speeded information processing or enhanced
attention.
In conclusion, our ndings have direct implications for cognitive
neuroscience studies involving tDCS, in that they suggest that investigators can guide the effects of stimulation by choosing
appropriate concurrent tasks, thereby enhancing the cognitive
resolution of the technique. Our ndings are also relevant to
potential clinical applications of tDCS, since they reinforce the
notion that pairing appropriate behavioral therapies with stimulation may be more effective than applying those therapies or tDCS
separately. Thus, in addition to parameters like intensity, duration,
electrode placement, and electrode polarity, task-dependent effects
may emerge as an important property to consider in tDCS studies of

cognition; additional studies that further dene the role of taskdependency are clearly merited.
Acknowledgments
We would like to thank the University Scholars program at the
University of Pennsylvania e Center for Undergraduate Research for
funding this research. We would also like to thank Gabriella Garcia
and Olufunsho Faseyitan for their contribution in the implementation of the project.

References
[1] Fregni F, Boggio PS, Nitsche M, et al. Anodal transcranial direct current stimulation of prefrontal cortex enhances working memory. Exp Brain Res
2005;166(1):23e30.
[2] Sparing R, Dafotakis M, Meister IG, Thirugnanasambandam N, Fink GR.
Enhancing language performance with non-invasive brain stimulation e a
transcranial direct current stimulation study in healthy humans. Neuropsychologia 2008;46(1):261e8.
[3] Moos K, Vossel S, Weidner R, Sparing R, Fink GR. Modulation of top-down
control of visual attention by cathodal tDCS over right IPS. J Neurosci
2012;32(46):16360e8.
[4] Marshall L, Molle M, Hallschmid M, Born J. Transcranial direct current
stimulation during sleep improves declarative memory. J Neurosci
2004;24(44):9985e92.
[5] Iyer MB, Mattu U, Grafman J, Lomarev M, Sato S, Wassermann EM. Safety and
cognitive effect of frontal DC brain polarization in healthy individuals.
Neurology 2005;64(5):872e5.
[6] Batsikadze G, Moliadze V, Paulus W, Kuo MF, Nitsche MA. Partially nonlinear stimulation intensity-dependent effects of direct current stimulation
on
motor
cortex
excitability
in
humans.
J
Physiol
2013;591(7):1987e2000.
[7] Boggio PS, Ferrucci R, Rigonatti SP, et al. Effects of transcranial direct current
stimulation on working memory in patients with Parkinsons disease. J Neurol
Sci 2006;249(1):31e8.
[8] Dockery CA, Hueckel-Weng R, Birbaumer N, Plewnia C. Enhancement of
planning ability by transcranial direct current stimulation. J Neurosci
2009;29(22):7271e7.
[9] Andrews SC, Hoy KE, Enticott PG, Daskalakis ZJ, Fitzgerald PB. Improving
working memory: the effect of combining cognitive activity and anodal
transcranial direct current stimulation to the left dorsolateral prefrontal cortex. Brain Stimul 2011;4(2):84e9.
[10] Segrave RA, Arnold S, Hoy K, Fitzgerald PB. Concurrent cognitive control
training augments the antidepressant efcacy of tDCS: a pilot study. Brain
Stimul 2014;7(2):325e31.
[11] Ohn SH, Park CI, Yoo WK, et al. Time-dependent effect of transcranial direct
current stimulation on the enhancement of working memory. Neuroreport
2008;19(1):43e7.
[12] Kincses T, Antal A, Nitsche M, Bartfai O, Paulus W. Facilitation of probabilistic
classication learning by transcranial direct current stimulation of the prefrontal cortex in the human. Neuropsychologia 2003;42:113e7.
[13] Bikson M, Rahman A. Origins of specicity during tDCS: anatomical, activityselective, and input-bias mechanisms. Front Hum Neurosci 2013:7.
[14] DEsposito M, Aguirre GK, Zarahn E, Ballard D, Shin RK, Lease J. Functional MRI
studies of spatial and nonspatial working memory. Cogn Brain Res
1998;7(1):1e13.
[15] Mottaghy FM, Krause BJ, Kemna LJ, et al. Modulation of the neuronal circuitry
subserving working memory in healthy human subjects by repetitive transcranial magnetic stimulation. Neurosci Lett 2000;280(3):167e70.
[16] Nagel IE, Preuschhof C, Li SC, et al. Load modulation of BOLD response and
connectivity predicts working memory performance in younger and older
adults. J Cogn Neurosci 2011;23(8):2030e45.
[17] Filmer HL, Mattingley JB, Dux PE. Improved multitasking following prefrontal
tDCS. Cortex 2013;49(10):2845e52.
[18] Karim AA, Schneider M, Lotze M, Veit R, Sauseng P, Braun C, et al. The truth
about lying: inhibition of the anterior prefrontal cortex improves deceptive
behavior. Cereb Cortex 2010;20(1):205e13.
[19] Weiss M, Lavidor M. When less is more: evidence for a facilitative
cathodal tDCS effect in attentional abilities. J Cogn Neurosci
2012;24(9):1826e33.
[20] Tombaugh TN. A comprehensive review of the paced auditory serial addition
test (PASAT). Arch Clin Neuropsychol 2006;21(1):53e76.
[21] Heinzel S, Lorenz RC, Brockhaus WR, et al. Working memory load-dependent
brain response predicts behavioral training gains in older adults. J Neurosci
2014;34(4):1224e33.
[22] Nitsche MA, Paulus W. Excitability changes induced in the human motor
cortex by weak transcranial direct current stimulation. J Physiol
2000;527(3):633e9.

J. Gill et al. / Brain Stimulation 8 (2015) 253e259

[23] Nitsche MA, Paulus W. Sustained excitability elevations induced by transcranial
DC motor cortex stimulation in humans. Neurology 2001;57(10):1899e901.
[24] Datta A, Bansal V, Diaz J, Patel J, Reato D, Bikson M. Gyri-precise head model
of transcranial direct current stimulation: Improved spatial focality using
a ring electrode versus conventional rectangular pad. Brain Stimul
2009;2(4):201e7.
[25] Reis J, Schambra HM, Cohen LG, et al. Noninvasive cortical stimulation
enhances motor skill acquisition over multiple days through an effect on
consolidation. Proc Natl Acad Sci U S A 2009;106(5):1590e5.
[26] Meinzer M, Jaehnigen S, Copland DA, et al. Transcranial direct current stimulation over multiple days improves learning and maintenance of a novel
vocabulary. Cortex 2014;50:137e47.
[27] Martin DM, Liu R, Alonzo A, et al. Can transcranial direct current
stimulation enhance outcomes from cognitive training? A randomized

[28]

[29]
[30]

[31]

259

controlled trial in healthy participants. Int J Neuropsychopharmacol

2013;16(9):1927e36.
Richmond L, Wolk D, Olson I. Enhancing training and transfer effects by
applying non-invasive brain stimulation during working memory training.
J Cogn Neurosci 2013:193.
Gronwall DMA. Paced auditory serial-addition task e measure of recovery
from concussion. Percept Mot Skills 1977;44(2):367e73.
Lockwood AH, Linn RT, Szymanski H, Coad ML, Wack DS. Mapping the neural
systems that mediate the Paced Auditory Serial Addition Task (PASAT). J Int
Neuropsychol Soc 2004;10(1):26e34.
Royan J, Tombaugh TN, Rees L, Francis M. The Adjusting-Paced Serial Addition
Test (Adjusting-PSAT): thresholds for speed of information processing as a
function of stimulus modality and problem complexity. Arch Clin Neuropsychol 2004;19(1):131e43.