Accepted Manuscript
Effects of different electrical brain stimulation protocols on subcomponents of motorskill learning
George Prichard, Cornelius Weiller, Brita Fritsch, Janine Reis
PII: S1935-861X(14)00164-8
DOI: 10.1016/j.brs.2014.04.005
Reference: BRS 540
To appear in: Brain Stimulation
Received Date: 2 January 2014
Revised Date: 9 April 2014
Accepted Date: 11 April 2014
Please cite this article as: Prichard G, Weiller C, Fritsch B, Reis J, Effects of different electrical brainstimulation protocols on subcomponents of motor skill learning, Brain Stimulation (2014), doi: 10.1016/j.brs.2014.04.005.
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service toour customers we are providing this early version of the manuscript. The manuscript will undergocopyediting, typesetting, and review of the resulting proof before it is published in its final form. Pleasenote that during the production process errors may be discovered which could affect the content, and alllegal disclaimers that apply to the journal pertain.
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
1
Original Article
Effects of different electrical brain stimulation protocols on
subcomponents of motor skill learning
George Prichard1, 2, 3, Cornelius Weiller1, Brita Fritsch1*, Janine Reis1*
1Department of Neurology, Albert-Ludwigs-University Freiburg, Germany.
2Faculty of Behavioral and Social Sciences, University of Groningen, The Netherlands.
3 Institute of Cognitive Neuroscience, University College London, United Kingdom.
* These authors contributed equally
Corresponding Author: Janine Reis, [email protected], Dept of Neurology, University Hospital Freiburg, Breisacher Straße 64, 79106 Freiburg, Germany phone: +49 (0)761 270 50010 Fax: +49 (0)761 270 53900
Running Title: tDCS, tRNS and motor skill learning
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
2
Abstract
Background: Noninvasive electrical brain stimulation (NEBS) with
transcranial direct current (tDCS) or random noise stimulation (tRNS)
applied to the primary motor cortex (M1) can augment motor learning.
Objective: We tested whether different types of stimulation alter particular
aspects of learning a tracing task over three consecutive days, namely skill
acquisition (online/within session effects) or consolidation (offline/between
session effects).
Methods: Motor training on a tracing task over three consecutive days was
combined with different types and montages of stimulation (tDCS, tRNS).
Results: Unilateral M1 stimulation using tRNS as well as unilateral and
bilateral M1 tDCS all enhanced motor skill learning compared to sham
stimulation. In all groups, this appeared to be driven by online effects
without an additional offline effect. Unilateral tDCS resulted in large skill
gains immediately following the onset of stimulation, while tRNS exerted
more gradual effects. Control stimulation of the right temporal lobe did not
enhance skill learning relative to sham.
Conclusions: The mechanisms of action of tDCS and tRNS are likely
different. Hence, the time course of skill improvement within sessions could
point to specific and temporally distinct interactions with the physiological
process of motor skill learning. Exploring the parameters of NEBS on
different tasks and in patients with brain injury will allow us to maximise the
benefits of NEBS for neurorehabilitation.
Keywords: tDCS, tRNS, brain stimulation
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
3
Introduction
A substantial portion of our lives is spent learning new motor skills: from
walking to writing, driving and sports. Motor skills are the primary
mechanism for interaction with the world around us; hence, defective motor
skills resulting from neurological diseases are a severe impairment.
Noninvasive electrical brain stimulation (NEBS) applied transcranially to the
motor cortex (M1) has been shown to improve motor skill learning in
healthy individuals [1,2] and in chronic stroke patients [3–5]. Transcranial
direct current stimulation (tDCS) with an anode over M1 and a cathode over
the contralateral supraorbital area, in combination with motor training
resulted in greater skill gains compared to sham in healthy subjects. Using
tasks of different complexity, experiments have shown both within session
(online) improvements in a single-day [6,7], as well as between session
(offline) improvements observed with multi-session training [8,9]. In an
attempt to maximize stimulation benefits, recent studies utilized a bilateral
M1 montage, with an anode over the M1 contralateral and a cathode
ispilateral to the training hand. The basic idea of this approach is the
modulation of interhemispheric inhibition [10–12], that is strengthening the
facilitatory effect on one M1 with anodal tDCS, while reducing the inhibitory
influence of the other M1 by cathodal tDCS [12,13].
While tDCS uses a direct current flowing in one direction necessitating an
anode and cathode with potentially different local effects, transcranial
random noise stimulation (tRNS) uses an alternating current with a
randomly changing frequency and current direction, removing
anode/cathode-specific effects. High frequency tRNS (100-640Hz) applied
to M1 has also been shown to facilitate implicit motor sequence learning
[14]. Both anodal tDCS and tRNS enhance M1 excitability [14–16],
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
4
although it is likely there are differences in the mechanism of action of a
constant current versus changing currents applied to the cortex [17,18]. It
has been suggested that the concept of stochastic resonance may apply to
all forms of NEBS: stimulation-induced noise introduced to a neuronal
system may provide a signal processing benefit in the brain by altering the
signal-to-noise ratio [19,20]. While synchronization with task-relevant
activity may play a particular role for tRNS, additional homeostatic
mechanisms induced by a constant noise input may apply for tDCS [19,21].
Despite the huge amount of separate investigations of tDCS and tRNS
effects assessed in a single session, there are only two direct comparisons
between these stimulation types: For visuomotor learning, neither tRNS nor
anodal tDCS applied to M1 combined with a brief single training session
improved learning relative to sham stimulation [22]. On an orientation
discrimination task Pirulli et al. [23] found the effect of stimulation type
applied to the visual cortex varied depending on the timing of stimulation,
with tRNS more effective if applied during practice, whereas tDCS induced
better discrimination when applied before practice. However, these results
directly contrast with results from the motor learning domain, where anodal
tDCS applied to M1 before learning a serial reaction time task was found to
inhibit or leave unaffected subsequent learning [7,24], showing NEBS
effects are current type, site and task specific.
Given that these studies probed aspects of learning in a single session and
one was not directly related to motor learning, it is currently unknown
whether tDCS and tRNS would exert distinct effects onto specific
subcomponents of motor skill learning, i.e. within session (online)
improvements or between session (offline) effects, only assessable when
training for more than one session. Disentangling how different forms of
NEBS interact with the stages of the learning process is of great value both
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
5
for understanding the mechanisms of motor skill learning as well as to
maximize clinical benefits of NEBS. Here, we directly contrast the effects of
tDCS and tRNS on repeated motor learning sessions in an exploratory
study to test the efficacy of these different stimulation types.
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
6
Methods and Materials
This study was in accordance with the Declaration of Helsinki amended by
the 59th WMA General Assembly, Seoul, October 2008 and was approved
by the local Ethics Committee of the University of Freiburg.
Participants
Subjects were invited for participation by bulletin board announcements at
the university as well as by word-of-mouth and social media. 91 healthy,
adult participants (39 males, mean age = 25.7, SD = 4.6 years) were
recruited to the experiment and given a small monetary reimbursement. All
participants gave written informed consent, and met safety criteria for TMS
and tDCS [25]. All were right handed as assessed by the Edinburgh
Handedness Inventory (mean score = 93.4, SD = 10.7). Inclusion required
a normal neurological or psychiatric medical history. Subjects were also
screened for symptoms of depression using the Beck Depression Inventory
([26], score > 12 leading to exclusion). In addition, all subjects were
screened for the brain-derived neurotrophic factor (BDNF) val66met
polymorphism known to affect motor skill learning [27] and the distribution
of genotypes (val66val, Met carriers) within each stimulation group was
monitored to avoid a confounding effect in a particular stimulation group.
Subjects were not excluded due to their genotype.
The Tracing Task
We created a complex, continuous word/shape tracing task with the
Psychophysics Toolbox package [28,29] for Mathworks MATLAB, on the
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
7
basis of established hand tracing tasks [30–32]. Participants were required
to trace over a series of words and shapes on a Wacom Bamboo Fun
graphics tablet with a stylus held in the non-dominant (left) hand. The non-
dominant hand was selected in order to increase task difficulty for a steeper
learning curve. The template letters were presented separately on a
monitor (Figure 1A and 1B). Participants could move the cursor to a ready
position by moving the stylus without touching the tablet; touching initiated
the trial and started a time bar showing how long was left for the trial. Each
trial allowed 2 seconds per real letter or shape letter (e.g. 10 seconds for a
5 letter word). This timing was selected from speed accuracy trade-off data
from a behavioural pilot, finding 2 seconds per letter difficult but not
frustrating. Instructions were to trace as accurately as possible over the
template using all of the given time and lift the stylus on finishing a trial,
allowing us to measure the time taken per trial.
Template words consisted of the most common 3-5 letter German words
selected from a free database (compiled from subtitles by Invoke IT
Limited, http://invokeit.wordpress.com/frequency-word-lists/) which were
screened to remove emotionally salient words. Words were printed in a
freely available cursive font (League Script,
www.theleagueofmoveabletype.com). For template shapes, an alphabet of
random shapes was drawn in Inkscape (http://www.inkscape.org/), where
each shape corresponded to each letter of the alphabet. Using this, each
real word was converted into a ‘shape word’, where each real letter was
replaced by a made-up shape. For instance, the word ‘der’ (Figure 1A) was
turned into a shape-word with shape-letters corresponding to each real
letter (Figure 1B). A single trial consisted of tracing over one template (one
real word or one shape-word).
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
8
- Figure 1 near here -
Scoring Method
To measure participants’ performance, we devised a scoring method which
allowed for intuitive feedback and analysis. In brief, the final score can be
interpreted as 'percentage correct': a participant’s trace which perfectly
matches the template receives a score of 100; any deviation from this
(drawing off the template lines) reduces the score. Drawing very little or
consistently far away from the target lines results in a score at or
approaching zero.
In order to calculate these scores, both the trace and template (target) data
was converted into an image (Figure 1 C and D). The sum of the
differences between the two images was used as scoring method (Figure
1E). In order to introduce a margin of error, both images were blurred with a
Gaussian kernel (size: 50x50 pixels; standard deviation: 12 pixels; Figure 1
C and D) - this allows minor deviations, and makes the score worse the
further the trace deviates from the target, with a cut-off if deviations are too
far. At this point, a perfect trace gives 0 and deviations are arbitrarily high.
Therefore, we set the sum of the pixels in the template image as a
threshold upper score (e.g. writing nothing is the baseline for worst score).
The score was thresholded at this number, then divided by it and
subtracted from 1. This gives a fraction (with 0 as the worst score and 1 as
the best), which we turned into a percentage.
Study design
The study design is shown in Figure 2. We chose a parallel study design.
Subject allocation to one of the four stimulation conditions followed a fully
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
9
balanced randomization list prepared prior to the experiment. The electrode
montage was known to both participant and experimenter; however both
participants and the experimenter were blinded to the type (tRNS or tDCS)
and verity of stimulation (sham or not sham) by stimulation codes entered
into the electrical stimulator. All participants were naïve to the task and all
practiced the same version of the task, except for the fact that the order of
words/shape words within each training block was automatically pseudo-
randomized.
- Figure 2 near here -
Participants practiced the tracing task over 3 consecutive days at the same
time of day (+/- 1 hour). After locating the motor cortex using TMS on day 1,
the position was marked with a skin safe marker for use on day 2. NEBS
electrodes were secured in the appropriate location using bandages. On
day 1 and 2 participants completed 12 blocks of 15 trials (180 trials total) of
the tracing task, with an enforced 15s break between each block where
they were shown their average score over the last block for motivational
purposes. Stimulation was switched on after completion of block 1. Training
usually lasted longer than 20 minutes (35-40 minutes); thus, training
continued after the offset of stimulation. This approach was chosen to
mimic previous study designs [7–9] for comparability, as well as for safety
reasons (no tRNS studies are published to our knowledge with stimulation
durations above 20 minutes). On day 3, participants completed 8 blocks
without stimulation to test for any delayed carry-over effects of stimulation.
Motor cortex localisation
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
10
For all participants transcranial magnetic stimulation (TMS) using a
Magstim 200 and a custom-made figure-of-8-shaped coil (outer diameter
7cm, all Magstim, UK) was used to locate the left and right motor cortex for
NEBS electrode placement (including participants in the control group). The
“hot spot” was defined as the best location to consistently elicit motor
evoked potentials (MEPs) in the contralateral first dorsal interosseus
muscle.
Noninvasive Electrical Brain Stimulation
All stimulations except sham consisted of 20 minutes of stimulation with 15
seconds at the beginning and end slowly ramping up or down the current to
avoid discomfort or phosphenes. In the sham group, current was ramped
up then down over 30 seconds and current type (tRNS or tDCS) was
randomly selected. All electrical stimulation was applied using a neuroConn
DC-MR stimulator (Ilmenau, Germany). For tDCS 1mA of direct current
(assumed current density 0.0625 mA/cm2) was applied using carbon
electrodes covered by a square 16cm2 sponge. High frequency tRNS was
applied as in [14]. Random current levels were generated for each sample
(sampling rate 1280Hz, maximum frequency 640Hz), and a digital high-
pass filter was applied at 100Hz, resulting in a maximal current density of
0.0625 mA/cm2.
Electrode positions are shown in Figure 2 (unilateral M1 stimulation for both
tDCS and tRNS, bilateral M1 stimulation for tDCS). A non-M1 tRNS
stimulation group was included to address regional specificity of tRNS on
motor skill learning. In this group the right electrode was placed on the
position T6 of the 10-20 EEG system (tRNS:T6-SO). The temporal-parietal-
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
11
occipital junction (T6) was selected as an area where current flow would
not directly affect motor areas [33].
Psychophysical assessment
Before stimulation on day 1 and after finishing the experiment on day 3
subjects completed the positive and negative affect scale (PANAS, [34]) in
order to detect any effects of stimulation on mood. Before and after each
session on every day, participants rated their mental fitness on a visual-
analogue scale (VAS) from 1-10, where 1 was defined as ‘perfect mental
fitness’ and 10 was ‘extremely mentally unfit’. Each day we recorded
potential side effects and sleep duration the previous night. At the end of
day 3 participants were asked to guess whether they received real or sham
stimulation in order to check whether the blinding procedure was effective.
Analysis of Subcomponents of Motor Skill Learning
Subcomponents of motor skill learning were assessed by taking score
differences at various time points. This allowed us to break overall learning
down into online and offline effects, with immediate effect as a separate
measure (see Figure 2).
Overall learning was defined as the score difference on the first block of
day 3. Online (within day) effects were defined as the sum of score
differences between the first and last block of days 1 and 2 (Day1Block12-
Day1Block1) + (Day2Block12-Day2Block1). Offline (between day) effects
were defined as the sum of the differences between the last and first blocks
of consecutive days (Day2Block1-Day1Block12) + (Day3Block1-
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
12
Day2Block12). Immediate effects – initial changes after stimulation – were
the sum of the differences between the first two blocks on stimulation days
(Day1Block2-Day1Block1) + (Day2Block2-Day2Block1). Overall effects
(used for graphs only) were scores from Day3Block1 – Day1Block1. Day 3
learning was defined as the last block of day 3 minus the first block
(Day3Block8-Day3Block1) and was used to see if stimulation caused any
non-transient changes in learning in the absence of continued stimulation.
Statistical analysis
To ensure that participants performed the task as instructed (keeping
movement speed at 2 seconds per letter), a repeated measures ANOVA
was performed on “movement speed” with factor GROUP (sham, tDCS:M1-
SO, tDCS:M1-M1 and tRNS:M1-SO) as the between-subject variable and
factor TIME as within-subject variable. This was needed to control for an
additional shift in the speed-accuracy trade-off, e.g. sacrificing accuracy for
higher speed.
An ANCOVA with accuracy skill score on Day3Block1 as the dependent
variable, factor group (sham, tDCS:M1-SO, tDCS:M1-M1 and tRNS:M1-
SO) as the independent variable and baseline skill score (Day1Block1) as a
covariate was performed to test for total (overall) learning. Separate
ANOVAs were performed for online, offline and immediate effects with the
score change as the dependent variable and factor group (sham, tDCS:M1-
SO, tDCS:M1-M1 and tRNS:M1-SO) as the independent variable. Another
ANCOVA tested total day one learning (Day1Block12 as dependent
variable, Day1Block1 as covariate) to allow for a direct comparison with
previous single-session NEBS studies; a further ANOVA tested day 3
learning for any lasting effects of stimulation on learning (Day3Block8-
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
13
Day3Block1). Post hoc pairwise t-tests were used to discern where
differences lay. Since the project was set up as an exploratory experiment,
this would not necessarily require multiple test adjustment. However, to
control for a potential type I error without increasing the probability of
creating a type II error, we predefined two independent hypotheses for the
main experiment tested with Bonferroni-Holm adjustment for multiple
comparisons [35]. We hypothesized that 1) all 3 stimulation types show
greater overall learning than sham (3 pairwise comparisons) and 2) the 3
stimulation types differ with regard to overall learning and learning
subcomponents (3 pairwise comparisons). For the tRNS:T6-SO control
experiment added a posteriori, an ANCOVA with score on Day3Block1 as
the dependent variable, factor group (sham, tRNS:T6-SO) as the
independent variable and baseline skill score (Day1Block1) as a covariate
was calculated. Subcomponents of learning were not assessed.
ANOVAs for demographical data (Edinburgh Handedness score, age) were
performed with group (sham, tDCS:M1-SO, tDCS:M1-M1, tRNS:M1-SO
and tRNS:T6-SO) as a factor. Repeated measures ANOVAs with factor
group and time were used to assess changes in sleep, PANAS and VAS
scores. Chi-squared tests were used to test for binary differences between
each group on condition guess, gender and val66met polymorphism.
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
14
Results
Demographics
91 healthy German-speaking participants took part in the experiment. One
participant was excluded from analysis (including demographics) due to an
error in stimulation parameters (tDCS:M1-SO group). No further
participants were excluded. All participants tolerated the stimulation well. 8
subjects out of 90 reported mild headache after stimulation on day 1 only;
this is likely to be caused by TMS rather than NEBS, as they did not
reoccur on day 2 (no TMS), and 2 of these 8 were in the sham group. 1
subject in the tRNS:M1-SO group reported a mild transient headache under
the SO electrode on day 2 which passed before stimulation ended.
No significant differences between groups were found for age or
handedness (Table 1).
There was a significant effect of time on negative PANAS (p<0.001) scores,
with participants showing lower scores (less depressed/negative) on both
measures at the end of the experiment. However, there was no significant
effect of group or group x time. Positive PANAS scores showed no
significant changes. Mental fitness VAS scores significantly decreased after
each session (p<0.001), however there was no significant difference
between groups. There were no significant effects of day, group or group x
time on sleep (Table 2).
Separate chi-square analyses showed no significant differences between
groups on BDNF val66met polymorphism, gender or condition guess (Table
1). Though not significantly different from other groups, participants who
received tRNS guessed they had been stimulated less often than subjects
receiving sham stimulation (Table 1).
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
15
Adherence to task demands
Within the allowed movement speed range (max. 2 sec per letter),
participants slightly slowed down their movement speed (from 1.89 sec per
letter on day 1 to 1.92 s per letter on day 3; 30 ms change on average) over
the course of the experiment (rmANOVA: p <0.0001 for factor TIME).
However, there was no significant effect of GROUP (p = 0.761) and no
interaction (p = 0.419). In order to assess whether this needed to be
accounted for when calculating the subjects’ accuracy score, we estimated
the percentage of variance in scores that was accounted for by time taken
(normalised to number of letters per trial) to be 2% (R2 = 0.020). This low
explained variance, and lack of a significant difference between groups
indicates that improving scores over time was not just due to participants
slowing down and/or moving along their initial speed-accuracy trade-off, so,
we did not apply any corrections for movement speed.
Motor Cortex Stimulation
Overall Learning: The ANCOVA on day 3 block 1 scores showed a
significant effect of group [F(3, 68) = 3.097, p = 0.033] with an effect size
(eta2) of 0.12 and an observed power of 70% (Figure 3). Post-hoc tests
(hypothesis 1) showed this was due to significantly higher scores achieved
in tRNS:M1-SO (p=0.024), tDCS:M1-M1 (p = 0.030), and tDCS:M1-SO (p
= 0.047) compared to sham. No significant differences in skill at d3Block1
were observed between the 3 different motoric stimulation groups
(hypothesis 2, all p-values between 0.8 and 1).
- Figure 3 near here –
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
16
Subcomponents of motor skill learning: In order to provide a direct
comparison with the majority of previous motor learning studies, an
ANCOVA was performed on day 1 online effects (block 12) with each
motoric stimulation group as a factor and day 1 block 1 as a covariate. This
was not significant [F(3,68) = 1.666, p = 0.183].
An ANOVA on online effects was not significant [F(3,68) = 1.068, p =
0.386], though Figure 3 suggests a trend; nor was there a significant
difference between stimulation groups for offline effects [F(3,68) = 0.164, p
= 0.920]. There was a significant effect of stimulation group on immediate
effects [F(3,68) = 4.054, p = 0.010], this showed a particularly large effect
size (eta2 = 0.152) and an observed power of 82%. Post-hoc t-tests
(hypothesis 1) revealed this was due to an increased immediate score for
tDCS:M1-SO compared to sham (p = 0.002). There was also a significant
difference between tDCS:M1-SO and tRNS:M1-SO (p= 0.013; Figure 3). An
ANOVA on day 3 learning (last block of day 3 minus first block of day 3)
was not significant [F(3,68) = 0.454, p = 0.715]; it does not appear that
electrical stimulation has online effects on days subsequent to stimulation.
Control experiment
Since all M1 stimulation types yielded a significant effect on overall motor
skill learning compared to sham and the montages utilized typically affect a
relatively wide cortical and subcortical network [36], we were concerned
that effects of NEBS may be regionally unspecific. We added a tRNS
stimulation group not targeting the motor cortex (tRNS:T6-SO) in order to
establish whether stimulation excluding M1 would also enhance overall skill
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
17
learning. An ANCOVA on skill achieved by the beginning of day 3 found no
significant difference between sham and tRNS:T6-SO stimulation [F(1,33) =
2.226, p = 0.15] (Figure 4).
- Figure 4 near here -
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
18
oDiscussion
Motor cortex electrical stimulation accelerates prolonged motor skill
learning with different tDCS montages as well as with tRNS. Enhancement
of overall learning is mainly driven by online (within session) improvements.
No significant differences are found between tDCS and tRNS. Non-M1
stimulation does not significantly enhance motor skill learning, confirming
the key role of the primary motor cortex in motor skill learning and region
specific effects of stimulation.
The Tracing Task
Previous NEBS studies used the implicit serial reaction time task
[6,14], or an explicit reaction time task [7], recording reaction time while
learning a sequence. In contrast, the tracing task requires complex,
synergistic and continuous movements of multiple hand and arm muscles,
as well as hand-eye coordination. It is similar to a visuomotor task we
introduced earlier - the sequential visual isometric pinch task (SVIPT, [8]):
for both tasks skill depends on both speed and movement accuracy. With
practicing the tracing task subjects improved their accuracy at one fixed
(predetermined) movement speed in the task’s speed-accuracy trade-off.
In contrast to the SVIPT, in which both factors were freely changeable by
the subject, improving accuracy at the cost of significantly slower speeds
(seconds) was not possible for the tracing task. As depicted by the learning
curve of sham stimulated subjects (Figure 3), skill on the tracing task is
continuously increased within training sessions (online), followed by a small
loss in skill between training days (offline). The small loss of skill between
sessions may represent a small warm-up decrement [31,37,38], since
subjects do not return to naïve levels of performance and the level of the
previous day is reinstated within the first 15 -30 trials of training.
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
19
Stimulation parameters
All types of M1 stimulation significantly increased motor skill
learning compared to sham, visible in Figure 3. This is the first
demonstration of improved motor skill learning over multiple days by
unilateral M1-tRNS and bilateral M1-tDCS. Surprisingly, from our data there
is no strong evidence for one type of active M1 stimulation being more
effective for enhancing overall motor skill learning than any other.
Subcomponents of Skill Learning
With regard to specific subcomponents of multi session motor skill
learning this is the first study directly comparing the effects of different
stimulation types. We observed a tendency (visible in Figure 3, upper) for
improved day 1 online effects for all these stimulations, which is in line with
previous single session studies using unilateral M1-tDCS [6,7,39], tRNS
[14] or bilateral M1-tDCS [40] on reaction time tasks. Although not
statistically significant, all types of M1 stimulation appeared to increase
online (within session) rather than offline (between session) effects. This
result is different from our previous findings when using the SVIPT [8,9]:
tDCS (as tDCS:M1-SO) enhanced motor skill learning over five days due to
offline/consolidation effects, which was a cumulative effect over several
days. There are several explanations due to task differences for this
discrepancy. First, skill on the tracing task may be more difficult to acquire
as there are multiple possible word/shape templates to trace, each
requiring the coordination of multiple muscle groups in the hand and arm,
rather than a single sequence executed with a small muscle group.
Second, subjects were potentially rewarded for improvements since a score
was presented after every training block [41]. Finally, the tracing task is
trained at a fixed high speed, whereas on the SVIPT both speed and
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
20
accuracy can be improved. It is therefore conceivable that unilateral M1-
tDCS may have differential online or offline effects on particular aspects of
motor tasks, e.g. either accuracy or speed. Indeed, task-dependency of
tDCS effects has been described for similar tasks [42].
A large immediate effect of stimulation was present in the tDCS:M1-
SO group visible as skill increase directly after the onset of stimulation in
block 2 on day 1 and 2 (Figure 3). This was clearly different from the
tRNS:M1-SO group. To our knowledge, an effect this fast has not been
mentioned previously, though may be visible in previous papers [6,8]. So is
the immediate effect a rapid acceleration of learning, an instantaneous
performance boost, or both? The immediate boost for tDCS:M1-SO could
be due to tDCS interacting with frontal networks involved in executive
function/working-memory [43,44], which tRNS and bilateral M1 tDCS may
not effect in the same way, due to different current parameters and
electrical field distributions, respectively. In favour of DC and RN currents
having opposing effects on frontal networks, visuomotor task-related
activity in the left frontal cortex was decreased after high frequency tRNS,
but not after tDCS [22]. Interestingly, the acceleration of learning rapidly
declined, resulting in DC:M1-SO having online/overall effects comparable
to other stimulation groups.
Potential mechanisms
Our results suggest that either tDCS:M1-SO has mechanistically
different effects to other forms of stimulation (particularly tRNS), or it may
affect the same learning mechanism over a different time course. tDCS has
been shown to rapidly affect MEPs after only a few minutes [15]; this
appears to be slightly more delayed for tRNS [45], though the two
stimulations have not been directly compared with regard to the
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
21
electrophysiological effects. Intuitively, one would assume that a constant
polarized current exerts different effects on the cellular and molecular level
than a frequently changing current. It has been suggested that the
polarization of neurons leading to subthreshold shifts in membrane
potential by tDCS may subsequently change neuronal firing rates and alter
spike timing which would then catalyse synaptic transmission and
behavioural effects [21,27,46,47]. Indeed, on a synaptic level, anodal tDCS
induces long-term potentiation when combined with a second input [27], a
mechanism that may be mimicked by training in the presence of anodal
tDCS in humans. Accordingly, behavioural improvements often cannot be
detected without training [27,48], suggesting an activity-dependent process
that requires synaptic co-activation. Terney et al. suggest tRNS acts by
repeatedly opening sodium channels causing membrane depolarisation
[14] as has been described for weak changing currents applied to neurons
in vitro [49,50]. To our knowledge, however, there is no direct research on
the cellular/molecular effects of tRNS. Hence, itis currently unknown what
neural mechanisms differentiate tDCS and tRNS. In comparison to tDCS, in
the conceptual framework of stochastic resonance, tRNS (i.e., externally
applied random noise) may have a higher likelihood to optimize task-
specific brain activity (endogenous noise) due to amplification of weak
signals [19,21]. Hence, it should be kept in mind that excitability alteration
may not be the only factor positively interacting with the physiological
process of motor skill learning.
Control Group
The non-M1 control group (tRNS:T6-SO) in which stimulation was
active though not over M1 showed no improvement in overall learning
compared to sham (Figure 4). T6 was chosen in order to minimise current
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
22
flow through motor areas, as used by Antal et al. [33], tRNS was chosen
because of better blinding of participants. Although the current flow or effect
of tRNS may be non-focal, spreading over more than the area directly
under the electrodes, the lack of effect of T6 stimulation supports the view
that stimulation effects are region specific.
Limitations
It should be noted that due to the size of the electrodes used and
the placement of the return electrode we cannot with complete certainty
attribute these learning improvements solely to M1; it is also likely that
motor areas adjacent to M1 or prefrontal areas are affected by stimulation,
contributing to our results. More focal brain stimulation techniques may help
in ensuring learning gains can be related to specific brain areas, for
instance high-definition tDCS [51] or repetitive TMS [52]. Moreover, the
nondominant (left) hand was used for training and the right M1 stimulated in
our study. Results may vary when training the dominant (right) hand, given
a predominant left-hemispheric representation for movements and motor
memories and potentially greater responsiveness of the left M1 to NEBS
[53,54].
Conclusion
NEBS applied to M1 enhances motor skill learning over multiple
sessions; on the tracing task this appears to be due to strengthening of skill
acquisition within session when training the nondominant hand. Strikingly,
the direct comparison of different stimulation protocols revealed no clear
advantage of a particular stimulation type or montage with regard to overall
learning. Hence, as long as M1 is targeted, current type and electrical field
distribution within the brain seem to have a subordinate role with regard to
the behavioural outcome. This information may be important for the design
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
23
of future clinically-oriented NEBS trials. Improvement of motor skill learning
induced by NEBS may also be effective for various other activities requiring
fine motor skills. Hence, the combination of NEBS and task-specific motor
training could be a promising approach to enhance general motor function
in patients with motor deficits, e.g. after stroke.
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
24
Acknowledgements
Thanks to Mark Nieuwenstein for comments on an earlier version of the
manuscript and to Michel Rijntjes for helpful comments on the task design.
Financial Disclosures
All authors of this manuscript report no biomedical financial interests or
potential conflicts of interest.
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
25
References
[1] Reis J, Fritsch B. Modulation of motor performance and motor learning by transcranial direct current stimulation. Curr Opin Neurol 2011;24:590–6.
[2] Reis J, Swayne OB, Vandermeeren Y, Camus M, Dimyan MA, Harris-Love M, et al. Contribution of transcranial magnetic stimulation to the understanding of cortical mechanisms involved in motor control. J Physiol 2008;586:325–51.
[3] Hummel FC, Cohen LG. Non-invasive brain stimulation: a new strategy to improve neurorehabilitation after stroke? Lancet Neurol 2006;5:708–12.
[4] Hummel F, Celnik P, Giraux P, Floel A, Wu WH, Gerloff C, et al. Effects of non-invasive cortical stimulation on skilled motor function in chronic stroke. Brain 2005;128:490–9.
[5] Lefebvre S, Laloux P, Peeters A, Desfontaines P, Jamart J, Vandermeeren Y. Dual-tDCS Enhances Online Motor Skill Learning and Long-Term Retention in Chronic Stroke Patients. Front Hum Neurosci 2012;6:343.
[6] Nitsche MA, Schauenburg A, Lang N, Liebetanz D, Exner C, Paulus W, et al. Facilitation of implicit motor learning by weak transcranial direct current stimulation of the primary motor cortex in the human. J Cogn Neurosci 2003;15:619–26.
[7] Stagg CJ, Jayaram G, Pastor D, Kincses ZT, Matthews PM, Johansen-Berg H. Polarity and timing-dependent effects of transcranial direct current stimulation in explicit motor learning. Neuropsychologia 2011;49:800–4.
[8] Reis J, Schambra HM, Cohen LG, Buch ER, Fritsch B, Zarahn E, 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;106:1590–5.
[9] Reis J, Fischer JT, Prichard G, Weiller C, Cohen LG, Fritsch B. Time- but Not Sleep-Dependent Consolidation of tDCS-Enhanced Visuomotor Skills. Cereb Cortex 2013:1–9.
[10] Ferbert A, Priori A, Rothwell JC, Day BL, Colebatch JG, Marsden CD. Interhemispheric inhibition of the human motor cortex. J Physiol 1992;453:525–46.
[11] Schambra HM, Sawaki L, Cohen LG. Modulation of excitability of human motor cortex (M1) by 1 Hz transcranial magnetic stimulation of the contralateral M1. Clin Neurophysiol 2003;114:130–3.
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
26
[12] Williams J a, Pascual-Leone A, Fregni F. Interhemispheric modulation induced by cortical stimulation and motor training. Phys Ther 2010;90:398–410.
[13] Kang EK, Paik NJ. Effect of a tDCS electrode montage on implicit motor sequence learning in healthy subjects. Exp Transl Stroke Med 2011;3:4.
[14] Terney D, Chaieb L, Moliadze V, Antal A, Paulus W. Increasing human brain excitability by transcranial high-frequency random noise stimulation. J Neurosci 2008;28:14147–55.
[15] Nitsche M, Paulus W. Excitability changes induced in the human motor cortex by weak transcranial direct current stimulation. J Physiol 2000;527 Pt 3:633–9.
[16] Nitsche MA, Paulus W. Sustained excitability elevations induced by transcranial DC motor cortex stimulation in humans. Neurology 2001;57:1899–901.
[17] Zaghi S, Acar M, Hultgren B, Boggio PS, Fregni F. Noninvasive brain stimulation with low-intensity electrical currents: putative mechanisms of action for direct and alternating current stimulation. Neuroscientist 2010;16:285–307.
[18] Schmidt S, Scholz M, Obermayer K, Brandt SA. Patterned Brain Stimulation, What a Framework with Rhythmic and Noisy Components Might Tell Us about Recovery Maximization. Front Hum Neurosci 2013;7:325.
[19] Collins JJ, Chow CC, Imhoff TT. Stochastic resonance without tuning. Nature 1995;376:236–8.
[20] Moss F, Ward LM, Sannita WG. Stochastic resonance and sensory information processing: a tutorial and review of application. Clin Neurophysiol 2004;115:267–81.
[21] Miniussi C, Harris J a, Ruzzoli M. Modelling non-invasive brain stimulation in cognitive neuroscience. Neurosci Biobehav Rev 2013;37:1702–12.
[22] Saiote C, Polanía R, Rosenberger K, Paulus W, Antal A. High-frequency TRNS reduces BOLD activity during visuomotor learning. PLoS One 2013;8:e59669.
[23] Pirulli C, Fertonani A, Miniussi C. The role of timing in the induction of neuromodulation in perceptual learning by transcranial electric stimulation. Brain Stimul 2013;6:683–9.
[24] Kuo M-F, Unger M, Liebetanz D, Lang N, Tergau F, Paulus W, et al. Limited impact of homeostatic plasticity on motor learning in humans. Neuropsychologia 2008;46:2122–8.
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
27
[25] Keel JC, Smith MJ, Wassermann EM. A safety screening questionnaire for transcranial magnetic stimulation. Clin Neurophysiol 2000;112:720.
[26] Beck AT, Steer RA. Internal consistencies of the original and revised Beck Depression Inventory. J Clin Psychol 1984;40:1365–7.
[27] Fritsch B, Reis J, Martinowich K, Schambra HM, Ji Y, Cohen LG, et al. Direct current stimulation promotes BDNF-dependent synaptic plasticity: potential implications for motor learning. Neuron 2010;66:198–204.
[28] Brainard DH. The psychophysics toolbox. Spat Vis 1997;10:433–6.
[29] Kleiner M, Brainard DH, Pelli DG, Ingling A, Murray R, Broussard C. What’s new in Psychtoolbox-3? Percept. 36, vol. 21, 2007, p. ECVP Abstract Supplement.
[30] Cohen MR. Individual and sex differences in speed of handwriting among high school students. Percept Mot Skills 1997;84:1428–30.
[31] Stratton SM, Liu Y-T, Hong SL, Mayer-Kress G, Newell KM. Snoddy (1926) revisited: time scales of motor learning. J Mot Behav 2007;39:503–15.
[32] Verschueren SM, Swinnen SP, Cordo PJ, Dounskaia N V. Proprioceptive control of multijoint movement: unimanual circle drawing. Exp Brain Res 1999;127:171–81.
[33] Antal A, Polania R, Schmidt-Samoa C, Dechent P, Paulus W. Transcranial direct current stimulation over the primary motor cortex during fMRI. Neuroimage 2011;55:590–6.
[34] Watson D, Clark LA, Tellegen A. Development and validation of brief measures of positive and negative affect: the PANAS scales. J Pers Soc Psychol 1988;54:1063–70.
[35] Bender R, Lange S. Adjusting for multiple testing--when and how? J Clin Epidemiol 2001;54:343–9.
[36] Saiote C, Turi Z, Paulus W, Antal A. Combining functional magnetic resonance imaging with transcranial electrical stimulation. Front Hum Neurosci 2013;7:435.
[37] Adams JA. The second facet of forgetting: a review of warmup decrement. Psychol Bull 1961;58:257–73.
[38] Schmidt RA, Lee TD. Motor Control and Learning: A Behavioral Emphasis. 2nd ed. Human Kinetics; 2005.
[39] Antal A, Nitsche MA, Kincses TZ, Kruse W, Hoffmann KP, Paulus W. Facilitation of visuo-motor learning by transcranial direct current stimulation of the motor and extrastriate visual areas in humans. Eur J Neurosci 2004;19:2888–92.
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
28
[40] Kang EK, Paik N-J. Effect of a tDCS electrode montage on implicit motor sequence learning in healthy subjects. Exp Transl Stroke Med 2011;3.
[41] Abe M, Schambra H, Wassermann EM, Luckenbaugh D, Schweighofer N, Cohen LG. Reward improves long-term retention of a motor memory through induction of offline memory gains. Curr Biol 2011;21:557–62.
[42] Saucedo Marquez CM, Zhang X, Swinnen SP, Meesen R, Wenderoth N. Task-specific effect of transcranial direct current stimulation on motor learning. Front Hum Neurosci 2013;7:333.
[43] Miller EK, Cohen JD. An integrative theory of prefrontal cortex function. Annu Rev Neurosci 2001;24:167–202.
[44] Kane MJ, Engle RW. The role of prefrontal cortex in working-memory capacity, executive attention, and general fluid intelligence: an individual-differences perspective. Psychon Bull Rev 2002;9:637–71.
[45] Chaieb L, Paulus W, Antal A. Evaluating aftereffects of short-duration transcranial random noise stimulation on cortical excitability. Neural Plast 2011;2011:105927.
[46] Radman T, Su Y, An JH, Parra LC, Bikson M. Spike timing amplifies the effect of electric fields on neurons: implications for endogenous field effects. J Neurosci 2007;27:3030–6.
[47] Bindman LJ, Lippold OCJ. The action of brief polarizing currents on the cerebral cortex of the rat ( 1 ) during current flow and ( 2 ) in the production of long-lasting after-effects. J Physiol 1964;172:369–82.
[48] Galea JM, Celnik PA. Brain polarization enhances the formation and retention of motor memories. J Neurophysiol 2009;102:294–301.
[49] Schoen I, Fromherz P. Extracellular stimulation of mammalian neurons through repetitive activation of Na+ channels by weak capacitive currents on a silicon chip. J Neurophysiol 2008;100:346–57.
[50] Bromm B. Die Natrium-Gleichrichtung der unterschwellig erregten Membran in der quantitativen Formulierung der Ionentheorie. Pflügers Arch 1968;302:233–44.
[51] Kuo H-I, Bikson M, Datta A, Minhas P, Paulus W, Kuo M-F, et al. Comparing cortical plasticity induced by conventional and high-definition 4 × 1 ring tDCS: A neurophysiological study. Brain Stimul 2012:1–5.
[52] Muellbacher W, Ziemann U, Wissel J, Dang N, Kofler M, Facchini S, et al. Early consolidation in human primary motor cortex. Nature 2002;415:640–4.
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
29
[53] Schambra HM, Abe M, Luckenbaugh DA, Reis J, Krakauer JW, Cohen LG. Probing for hemispheric specialization for motor skill learning: a transcranial direct current stimulation study. J Neurophysiol 2011;106:652–61.
[54] Schade S, Moliadze V, Paulus W, Antal A. Modulating neuronal excitability in the motor cortex with tDCS shows moderate hemispheric asymmetry due to subjects’ handedness: a pilot study. Restor Neurol Neurosci 2012;30:191–8.
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
30
Tables
Demographics 1
Table 1: Demographics for group size, age, Edinburgh Handedness Inventory, gender and BDNFval66met carriers. ANOVA scores for age, Edinburgh Handedness Inventory. ± shows standard deviation. Gender shows percentage of males per group, met carriers shows percentage per group with val66met polymorphism.
n Age ± Edinburgh
± Males (%)
Met carriers
(%)
Condition Guess
Sham 18 25.7 5.5 92.5 14.0 38.9 38.8 38.9
tDCS:M1-SO
18 24.7 3.1 90.6 9.4 50 44.4 55.6
tDCS:M1-M1
18 26.5 6.3 96.9 7.5 38.9 33.3 47.1
tRNS:M1-SO
18 26.0 3.4 93.1 11.3 44.4 22.2 33.3
tRNS:T6-SO 18 25.6 4.5 93.9 10.4 44.4 38.8 16.7
p 0.832 0.494 0.959 0.160
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
31
Demographics 2
Table 2: Values for PANAS, VAS and sleep scores at each collection point, as well as results of repeated measures ANOVAs. VAS: visual-analogue scale for
ratings on perceived mental fitness ranging from 1(perfect mental fitness) to 10 (extremely mentally unfit.)1
PANAS pre +
PANAS pre -
VAS 1 pre
VAS 1 post
Sleep 1 VAS 2 pre
VAS 2 post
Sleep 2 VAS 3 pre
VAS 3 post
Sleep 3 PANAS post +
PANAS post -
Sham 32.8 15.8 3.4 4.1 7.4 3.3 4.0 7.3 3.1 3.6 7.1 31.7 14.4
DC:M1-SO
31.8 14.9 2.9 4.3 7.1 2.9 3.6 7.2 2.7 3.3 7.2 31.6 13.4
DC:M1-M1
33.6 14.9 3.2 3.8 7.2 3.2 3.6 6.9 3.3 3.7 7.0 34.3 13.7
RN:M1-SO
32.2 15.6 2.8 3.4 7.1 3.3 3.8 6.9 3.2 3.4 6.8 31.4 13.4
RN:T6-SO
33.0 15.4 3.6 3.6 6.5 3.5 3.7 6.8 3.8 3.7 6.8 33.3 13.2
p (time) <0.001 0.882 0.552 <0.001
p (time* group)
0.237 0.907 0.523 0.802
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
32
Figure legends
Figure 1: The tracing task and scoring method. A and B: An example trial;
black lines show the template word (A) or shape (B). Red lines show the
trace. B shows the corresponding shape-word to the real word ‘der’. C: A
template shape after blurring (black lines in B). D: A traced shape after
blurring (red lines from B). E: The subtractive image (each pixel in C
subtracted from each pixel in D). Yellow shows pixels with a value of 0.
Blue and red show pixels which increase the raw score (decreasing the
intuitive score), where the template was not matched (blue) and the trace
deviated from the template (red).
Figure 2: Study design. Upper: Subjects practiced the tracing task for
three consecutive days. Electrode placements in each condition are shown
on the right side. Red = anode, green = cathode, blue = non-polarised
tRNS electrode. For sham stimulation electrode positions randomly varied
between the M1 stimulation conditions. . Adapted from Reis and Fritsch
(2011) with permission. Lower: training day and measurements of learning
subcomponents are shown. Day 1 and 2 consisted of 12 training blocks (15
words/shapes per block). Stimulation was started after block 1. Please refer
to methods for a description of learning subcomponents.
Figure 3: Upper: Improvement in skill score per stimulation group. From
stimulation onset the average score in each stimulation group is higher than
sham at every time point, showing the early and persistent effects of
stimulation. Data presented as mean, error bars represent S.E.M. Lower:
Separated components of motor learning. Scatter data points show
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
33
individual difference scores (points), mean (central line) and standard error
of the mean (bars) for each group. Improvements due to stimulation are
visible for overall learning. Plot shows overall measure as described in
methods; p-values calculated on ANCOVA with correction for baseline
performance. A trend is visible for online effects, no clear changes for
offline effects. Significant improvement in immediate effect is shown for
tDCS:M1-SO compared to sham and tRNS:M1-SO. Level of significance:
*p<0.05, **p<0.01. Data presented as mean, error bars represent S.E.M..
Figure 4: Non-M1 control stimulation. Improvement in skill score over time
for sham and tRNS:T6-SO control group is shown. There is no significant
difference in overall learning (measured in the ANCOVA for D3Block1).
Data presented as mean, error bars represent S.E.M.
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT