The primary motor cortex is associated with learning the absolute, but not
relative, timing dimension of a task: A tDCS study

Tércio Apolinário-Souza, Marco Aurélio Romano-Silva, Débora Mar-

ques de Miranda, Leandro Fernandes Malloy-Diniz, Rodolfo Novellino
Benda, Herbert Ugrinowitsch, Guilherme Menezes Lage

PII: S0031-9384(16)30112-3
DOI: doi: 10.1016/j.physbeh.2016.03.025
Reference: PHB 11262

To appear in: Physiology & Behavior

Received date: 23 December 2015

Revised date: 21 March 2016
Accepted date: 22 March 2016

Please cite this article as: Apolinário-Souza Tércio, Romano-Silva Marco Aurélio, de
Miranda Débora Marques, Malloy-Diniz Leandro Fernandes, Benda Rodolfo Novellino,
Ugrinowitsch Herbert, Lage Guilherme Menezes, The primary motor cortex is associated
with learning the absolute, but not relative, timing dimension of a task: A tDCS study,
Physiology & Behavior (2016), doi: 10.1016/j.physbeh.2016.03.025

This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.
 ACCEPTED MANUSCRIPT
1

The primary motor cortex is associated with learning the absolute, but not
relative, timing dimension of a task: A tDCS study

Tércio Apolinário-Souzaa,

T
IP
Marco Aurélio Romano-Silvab,

R
Débora Marques de Mirandab,

SC
Leandro Fernandes Malloy-Dinizb,

NU
Rodolfo Novellino Bendaa,
MA
Herbert Ugrinowitscha,
D

Guilherme Menezes Lagea, b*

TE

a
School of Physical Education, Physiotherapy and Occupational Therapy,
P

GEDAM, Universidade Federal de Minas Gerais, Belo Horizonte, Brazil.

CE

b
Faculty of Medicine, INCT de Medicina Molecular, Universidade Federal de
AC

Minas Gerais, Belo Horizonte, Brazil.

*Corresponding author:

Guilherme Menezes Lage

Rua Tenente Brito Melo, 516, ap.901. Barro Preto. Belo Horizonte – MG. CEP

30180-070. Brazil. Phone: 55-31-988840411. E-mail: menezeslage@gmail.com

 ACCEPTED MANUSCRIPT
2

Abstract

The functional role of the primary motor cortex (M1) in the production of

movement parameters, such as length, direction and force, is well known;

T
however, whether M1 is associated with the parametric adjustments in the

IP
absolute timing dimension of the task remains unknown. Previous studies have

R
not applied tasks and analyses that could separate the absolute (variant) and

SC
relative (invariant) dimensions. We applied transcranial direct current

NU
stimulation (tDCS) to M1 before motor practice to facilitate motor learning. A

sequential key-pressing task was practiced with two goals: learning the relative
MA
timing dimension and learning the absolute timing dimension. All effects of the

stimulation of M1 were observed only in the absolute dimension of the task.

D

Mainly, the stimulation was associated with better performance in the transfer
TE

test in the absolute dimension. Taken together, our results indicate that M1 is
P

an important area for learning the absolute timing dimension of a motor

CE

sequence.
AC

Keywords: Primary motor cortex (M1); Motor learning; Transcranial direct

current stimulation (tDCS); Fast learning; Slow learning; Offline learning.

 ACCEPTED MANUSCRIPT
3

1. Introduction

T
Two important features involved in the acquisition of skilled behavior are

IP
the consistency of performance and the adaptability to variations imposed by

R
environmental contexts [1]. Motor consistency refers to the capacity to produce

SC
a well-defined spatio-temporal pattern over time. Motor adaptability refers to the

ability to cause small fluctuations in the spatio-temporal pattern without losing

NU
its identity [2]. For example, when we observe a sequence of overhand
MA
volleyball serves, the spatio-temporal pattern of the movement’s components is

maintained over the trials (i.e., the proper technique is assured). However, this

movement pattern has a reduced ability to adapt to the dynamics of the

D
TE

environment (Fig. 1).

P

INSERT FIGURE 1 HERE

CE

Relative and absolute timing errors provide information concerning the

AC

improvement of motor consistency and motor adaptability through practice.

Decreasing relative timing errors during the course of practice is associated with

the acquisition of motor consistency, which is observed, for example, in the

temporal invariance of the components of a determined pattern of movement

(e.g., a generalized motor program). Conversely, diminishing absolute timing

errors during the practice reflects an increased adaptability to adjust the

duration of the pattern of movement [3]. Fine-tuning this movement parameter

allows for a specific pattern of movement to be adapted to the demands of a

particular situation.
 ACCEPTED MANUSCRIPT
4

Initially, the absolute and relative timing dimensions of a sequential skill

do not appear to be equally associated with the primary motor cortex (M1)

function. While the movement sequencing (movement pattern) is selected,

T
retrieved and organized by the premotor cortex and supplementary motor area

IP
[4], M1 may make a functional contribution to the production of movement

R
parameters, such as length, direction, force and the time derivative of force [5–

SC
7]. Nevertheless, studies investigating the role of M1 in motor behavior [6–8]

have not explored the role of M1 in the absolute timing dimension of a task or

NU
utilized tasks and analyses that would separate these two dimensions of the
MA
motor skill. Moreover, these studies have focused only on the relationships

between neural substrates and motor control. Previous studies of motor

learning have shown that M1 inhibition by transcranial magnetic stimulation

D
TE

affects movement timing learning but does not disturb the learning of movement

amplitude (spatial parameter) [9]. Nevertheless, Lin et al. [9] did not separately
P

investigate the absolute and relative timing dimensions.

CE

Our hypothesis is that M1 is strongly associated with the learning of

AC

absolute timing parameterization in a sequential motor task. Conversely, we

expect a weak association between the relative dimension of the task and M1

because pre-motor areas are more involved in the programming of a movement

sequence. To investigate these associations, we compared the effects of the

motor practice plus transcranial direct current stimulation (tDCS) over M1 with

the effects of the motor practice plus sham stimulation. Anodal tDCS over M1

facilitates motor performance and learning [10,11], most likely by eliciting long-

lasting, polarity-dependent changes in cortex excitability [7,12]. tDCS might

strengthen synaptic connections through a mechanism that is similar to long-

 ACCEPTED MANUSCRIPT
5

term potentiation (LTP), a cellular mechanism that underlies memory

consolidation and learning [13,14].

Two memory processes may be involved in motor learning and

T
adaptation: a fast process that quickly adapts but also decays quickly, and a

IP
slow process that adapts and decays gradually [15,16]. Each process is

R
updated based on the error produced from the previous trial during the practice,

SC
weighted by a learning rate, and each state decays over trials but at different

NU
paces. While the slow process (slow learning) responds weakly to error but

contributes to long-term retention, the fast process (fast learning) responds

MA
strongly to initial errors but has poor retention [15]. To our knowledge, there are

no studies investigating the fast and slow processes in the relative and absolute
D

dimension of the task. However, there is no reason to assume that these

TE

processes occur over the same temporal scale in each dimension of the task. If
P

tDCS enhances the mechanism that underlies memory formation [13], we can
CE

hypothesize that the stimulation of M1 will produce more pronounced slow

learning and, consequently, better learning of the absolute dimension of a task.

AC

We also expect that independent of the transcranial stimulation, the fast

learning will not directly interfere in skill consolidation.

Additionally, if M1 is more associated with the learning of the absolute

dimension of a task, improvements post-practice in the absolute dimension

should be more pronounced in the anodal-tDCS group (G_M1) than in the

sham-tDCS group (G_SH). Conversely, the level of gains of the relative

dimension should be similar between groups during practice and post-practice.

Improvement during practice is evaluated through skill modifications within a

 ACCEPTED MANUSCRIPT
6

session that is usually finished when an asymptotic performance is reached;

this improvement is frequently defined as online change [17]. The results of

Cantarero et al. [18] and Reis et al. [19] show no differences in the online

T
change between the stimulated and sham groups. Improvement post-practice,

IP
which is often defined as offline change, is assessed by the modifications in the

R
skill performance after a break period [13]. The results of Cantarero et al. [18]

SC
and Reis et al. [19] show that a tDCS effect is clearly observed in the offline

change. An initially fragile memory formed during the online change is

NU
transformed into a stable and endurable form during the offline change in a
MA
process termed motor memory consolidation [20–22]. While online change is

measured by the difference observed between the final and initial parts of the

practice, offline change is measured by the difference between the levels of

D
TE

performance achieved after a break period and in the final part of the practice

[19].
P
CE

We also expect better performance in the absolute dimension in the

retention and transfer tests in the G_M1 group, indicating better learning. Motor
AC

learning, which is defined as a set of processes associated with practice and

experience producing relatively permanent changes in the competence to

respond, can be evaluated over the period of practice through the subject’s

ability to retrieve the skill practiced, as measured by the retention test, or

transfer what has been learned to new task variants, as measured by the

transfer test [23–25].

Finally, as we applied tDCS prior to motor practice, we asked whether we

could expect differences in the initial performance within the practice session. A
 ACCEPTED MANUSCRIPT
7

few studies have used repeated measures to investigate the effects of

stimulation throughout practice [26,27]. These studies did not find differences

between the tDCS and sham groups in the initial performance of practice.

T
Behavioral effects of the tDCS do not directly mirror the robust

IP
electrophysiological effects (e.g., higher cortical excitability) that outlast the

R
stimulation period by up to 90 minutes [27]. If tDCS are time-dependent and

SC
strengthen synaptic connections through a mechanism that is similar to LTP, we

could expect that the effects of tDCS will be clearly observed in the analyses

involving changes after practice.

NU
MA
2. Methods

2.1. Participants
D
TE

Thirty-two undergraduate students (16 men and 16 women,

counterbalanced by groups), ranging from 18 to 35 years old (mean 26.4 ± 3.77

P

in the tDCS group and 25.3 ± 3.94 in the sham group), participated in this
CE

experiment. All participants were right-handed university students (mean

AC

laterality quotient in the Edinburgh Handedness Inventory [28] = 84.9 in the

tDCS group and 85.8 in the sham group) and had normal or corrected-to-normal

visual acuity in both eyes.

The volunteers had no prior experience with the experimental task and

were blinded to the type of experimental manipulation (tDCS or sham). A local

ethics committee approved all procedures, and subjects provided informed

consent after receiving a full explanation of the study.

2.2. Apparatus
 ACCEPTED MANUSCRIPT
8

A computer, color monitor, and numeric keypad were placed on a

standard table in the lab room. A custom-made software program was used to

control the experimental task and to register the time between pressing the keys

T
[2]. Participants were asked to sit on a chair in front of the computer monitor

IP
and to adjust the numeric keypad position to comfortably use it with their right

R
hand. tDCS (HDC Kit, Magstim Co., Whitland, UK) was used to stimulate the

SC
primary motor cortex (M1).

NU
2.2.1. tDCS

The HDC Kit (Magstim Co., Whitland, UK) induced a current (1 mA)
MA
through saline-soaked sponge electrodes (surface area of 25 cm 2). The current

was initially increased in a ramp-like fashion over 30 seconds until reaching 1

D

mA (current density of 0.04 mA/cm2). Constant current flow was controlled by

TE

our own device. To stimulate the left primary motor cortex (Fig. 2), the anodal
P

electrode was placed over C3 (international 10-20 system), and the cathodal
CE

electrode was placed above the Fp2 (right supraorbital). The tDCS was

delivered for 20 min before the motor task. The experimenter was not blinded to
AC

the type of experimental manipulation (tDCS or sham). The motor task was

executed immediately after stimulation.

In the sham group, stimulation was started in a ramp-like fashion but

faded out slowly after 36 seconds, which has been demonstrated to

successfully blind participants to the group [29]. The participants who received

the sham stimulation remained with the apparatus over their scalp throughout

the 20 min.

INSERT FIGURE 2 HERE

 ACCEPTED MANUSCRIPT
9

2.2.2. Task

The participants were asked to sequentially press four keys (2, 8, 6, and

4) on the numeric keypad using the index finger of their right hand, which was

T
maintained in a constant position during all phases of the experiment. The goals

IP
involved in the task execution, the three relative times between each key strike

R
and the movement time (MT) were presented on the computer screen before

SC
each trial. The three relative times between each key strike in all experimental

NU
phases were (1) 22.2% (key 2 to 8), (2) 44.4% (key 8 to 6) and (3) 33.3% (key 6

to 4). The MT was 900 ms for both groups (Fig. 3). The relative times between
MA
each key strike refers to the relative timing dimension of the task, while MT

represents the absolute dimension. After completing each trial, the knowledge
D

of the results (KR) was displayed on the screen. The KR included the relative
TE

time between each key strike, the relative error (RE) of the sum of the relative

times, and the total movement time performed.

P
CE

INSERT FIGURE 3 HERE

AC

2.3. Procedures

Detailed instructions were provided to each participant concerning the

information displayed on the computer screen, and all subjects were asked to

be as accurate as possible regarding both the relative and MT criteria. Before

each trial, the MT and the criterion segment ratios for the task appeared on the

screen. After the movement sequence was completed, the KR was displayed

for 6 seconds. If a participant pressed the keys in an incorrect order, a warning

was displayed and the trial was repeated. The participants were randomly
 ACCEPTED MANUSCRIPT
10

assigned to one of two practice groups: 1) constant with tDCS (G_M1) or 2)

constant with sham stimulation (G_SH) 3).

The experiment consisted of three parts: the practice (acquisition) phase,

T
the retention test and the transfer test. During practice, all participants

IP
performed 120 trials with the same MT and criterion segment ratios. The G_M1

R
group received stimulation before practice, and the G_SH group received sham

SC
stimulation.

The retention test was conducted 24 hours after participants practiced in

NU
a 12-trial block. The test utilized the same MT and relative criterion segment
MA
ratios as the practice test. The transfer test was performed immediately after the

retention test in a 12-trial block with a novel MT criterion (1,300 ms) and with

the same relative criterion segment ratios (22.2%, 44.4%, and 33.3%) as the
D
TE

practice phase. The KR and stimulation were not provided in the retention and

transfer tests.
P
CE

2.4. Measurements

Two dependent variables, RE and AE, were measured. RE was used as

AC

a measure of proficiency in the relative dimension, while AE was used as a

measure of proficiency in the absolute dimension. The RE was determined as

the sum of absolute differences between the observed and criterion time ratios

for each segment and was computed as follows:

RE= ΙR1 – 22.2Ι + ΙR2– 44.4Ι + ΙR3 – 33.3Ι, where R1, R2 and R3 = (the

actual movement time of segment / total movement time) X 100.

The AE was computed as the difference between the actual movement

time and the total criterion time:

 ACCEPTED MANUSCRIPT
11

AE= (MTn – total criterion time).

We used the model multi-rate learning process [15] to calculate fast and

slow learning (Fig. 4) in the absolute dimension. The four parameters used to

T
simulate the model (Af, Bf, As, and Bs) were obtained by correlating the MT and

IP
the mean motor output by group in each block of practice. The motor output

R
corresponded to the sum of the slow and fast processes. The following rules

SC
defined by Smith et al. [15] were applied to the model of multi-rate learning: (1)

NU
the four parameters should be positive; (2) the value of Bf must be several-fold

larger than Bs; (3) As must be several times closer to one than Af; and (4) the
MA
correlation value between MT and the mean motor output should exceed 0.9.

The parameters obtained for the G_M1 group were as follows: Af = 0.01, Bf =
D

0.002, As = 0.6, and Bs = 0.02. The parameters obtained for the G_SH group
TE

were as follows: Af = 0.01, Bf = 0.9, As = 0.03, and Bs = 0.05.

P

INSERT FIGURE 4 HERE

CE

After calculating the slow and fast learning values, we analyzed the
AC

correlations between the memory processes (fast and slow learning) and the

performance of the retention and transfer tests of each group. For this, we

estimated the behavior between the last block of the practice phase and the

retention and transfer tests. The amount of trials estimated was the same as

that used to measure the fast and slow learning during the practice. The

estimated curve was obtained from the intercept and slope values of the linear

regression between the last block of practice and the retention and transfer

tests. Then, two independent t tests were conducted to compare the level of fast
 ACCEPTED MANUSCRIPT
12

learning between the G_M1 and the G_SH groups and to compare the level of

slow learning.

The online and offline changes were calculated for both dependent

T
measures. The online change was calculated as the sum of the difference

IP
between the last block and the first block of the practice phase. The offline

R
change was calculated as the sum of the difference between the block of the

SC
retention test and the last block of the practice phase. Differences between the

NU
performance of the last block of the practice phase and the transfer test were

also calculated. This change in performance could not be defined as an offline

MA
change because there was an interference due to the retention test between the

practice phase and the transfer test.

D
TE

2.5. Data analyses

P

The data were organized into blocks of 12 trials. The performance in the
CE

practice phase was analyzed by a two-way ANOVA (2 groups X 10 blocks) with

repeated measures in the second factor. All post-hoc analyses were conducted
AC

using Tukey’s test. Data from the retention and transfer tests were analyzed

using independent t-tests or Mann-Whitney U tests (when the data were

nonparametric). Independent t-tests were conducted to compare the levels of

group correlations between (1) fast learning and the retention test; (2) fast

learning and the transfer test; (3) slow learning and the retention test; and (4)

slow learning and the transfer test. The online and offline effects and the

change between the last block of acquisition and the transfer test were

assessed using t-tests. An alpha level of .05 was chosen for all inferential

statistics. The effect sizes were calculated using eta-squared (ƞ²) for analyses
 ACCEPTED MANUSCRIPT
13

of variance, Cohen’s (d) for t-tests, and Cliff's Delta (δ) for Mann-Whitney U

tests.

3. Results

T
IP
3.1. Absolute dimension analyses

R
3.1.1. Absolute timing error (AE) during practice

SC
Descriptive statistics are presented in Figures 5a and 5e. The inferential

NU
analysis did not detect a main effect of the group condition (F(1,30) = 0.06, P >
MA
0.05, ƞ² = 0.05). An analysis of the block condition (F(9, 30) = 22.41, P < 0.001,

ƞ² = 1.0) followed by a post hoc analysis indicated that the errors were greater

in block one than in the other blocks (P < 0.001 for all comparisons). The group
D
TE

X block interaction was not significant (F(9,30) = 1.80, P > 0.05, ƞ² = 0.80).
P

3.1.2. AE in the retention and transfer tests

CE

Descriptive statistics are presented in Figure 5a. Although the inferential

AC

analysis did not detect differences between groups in the retention test (Z(32) =

0.42, P > 0.05, δ = 0.09), the transfer test analysis indicated that a significant

difference existed between groups (t(30) = -2.24, P < 0.05, d = 0.74). The

G_M1 group was more accurate than the G_SH group (Fig. 5a).

3.1.3. Fast and slow learning

Descriptive statistics are presented in Figures 5c and 5d. The inferential

analysis indicated that the levels of correlation between fast learning and the

retention test were identical between groups (t(30) = -0.83, P > 0.05, d = 0.29).
 ACCEPTED MANUSCRIPT
14

The same pattern of results was found in the levels of correlation between slow

learning and the retention test (t(30) = -1.56, P > 0.05, d = 0.55) and between

fast learning and the transfer test (t(30) = -0.02, P > 0.05, d = 0.007).

T
Nevertheless, the level of correlation between slow learning and the transfer

IP
test was higher in the G_M1 group than in the G_SH group (t(30) = -281.94, P >

R
0.001, d = 0.80).

SC
3.1.4. Online change, offline change and change between practice and transfer

NU
test MA
Descriptive statistics are presented in Figure 5b. The inferential analysis

of the online change demonstrated a difference between groups (Z(32) = -1.99,

P < 0.05, δ = 0.41). The G_M1 group exhibited a greater decrease in errors
D
TE

than the G_SH group (Fig. 5b). Although the analysis of the offline change did

not indicate a difference between groups (t(30) = 0.02, P > 0.05, d = 0.008), the
P

analysis of the change between the end of practice and the transfer test
CE

indicated a difference between groups (t(30) = -2.39, P < 0.05, d = 0.88) (Fig.
AC

2C). The G_SH group showed a greater increase in the error compared with the

G_M1 group during the offline learning (Fig. 5b).

INSERT FIGURE 5 HERE

3.2. Relative dimension analyses

3.2.1. Relative timing error (RE) during practice

Descriptive statistics are presented in Figures 6a and 6d. The inferential

analysis did not detect the main effect of the group condition (F(1, 30) = 0.55, P
 ACCEPTED MANUSCRIPT
15

> 0.05, ƞ² = 0.11). The analysis of the block condition (F(9, 30) = 32.20, P <

0.001, ƞ² = 1.0) followed by the post hoc analysis indicated that the errors were

greater in block one than in the other blocks (P > 0.05 for all comparisons). The

T
group X block interaction (F(9, 30) = 3.44, P < 0.001, ƞ² = 0.98) followed by the

IP
post hoc analysis indicated that errors were greater in block one than in the

R
other blocks of the G_M1 group (P < 0.05). Identical results were found for the

SC
G_SH group.

NU
3.2.2. RE in the retention and transfer tests
MA
Descriptive statistics are presented in Figure 6a. The inferential analysis

did not detect a difference between the groups in either the retention test (Z(32)

= 0.26, P > 0.05, δ = 0.05) or the transfer test (t(30) = -1.14, P > 0.05, d = 0.40).
D
TE

3.2.3. Online change, offline change and change between practice and transfer
P

tests
CE

Descriptive statistics are presented in Figure 6b. The inferential analysis

AC

did not detect a difference between groups for online change, Z(32) = -1.20, P >

0.05, δ = 0.25, for offline change, Z(32) = -1.01, P > 0.05, δ = 0.21, or for

change between the practice phase and the transfer test, t(30) = -1.39, P >

0.05, d = 0.27.

3.2.4. Relative timing among each movement component

Descriptive statistics for the relative timing between each movement

component (key 2 to 8; key 8 to 6; key 6 to 4) are presented in Figures 6e.

Throughout the practice, the second component in both groups indicated better
 ACCEPTED MANUSCRIPT
16

adjustment, achieving a 44% ratio of the total movement time.

INSERT FIGURE 6 HERE

T
4. Discussion

IP
Overall, the results partially confirmed our hypotheses. In several

R
analyses, the M1 stimulation was associated with the absolute dimension of the

SC
task. Specifically, (1) the slow learning process was associated with the

NU
performance in the transfer test only in G_M1; (2) M1 stimulation was

associated with a higher online change and a lower change between the
MA
practice and the transfer tests; and (3) G_M1 showed a better performance

compared to G_Sham in the transfer test. As expected, all analyses of the

D

relative dimension were identical between groups. We also hypothesized that

TE

the effects of tDCS would be clearly observed only in the analyses that involved
P

changes after practice. Although the performances of both groups were similar
CE

during the practice, the online change was more pronounced in the G_M1

group. M1 is an extensively studied cortical target in tDCS studies [30–32], and

AC

it is known that motor practice is associated with tDCS and facilitates motor

learning [11]; however, to our knowledge, this dimension-dependent effect on

M1 stimulation is a new finding.

As expected, we did not find differences in the initial performances of the

groups in both dimensions. Previous studies [25,26, exp.1 e 3, 33] that

stimulated M1 prior to practice also failed to find an initial difference between

the stimulated and sham groups. Altogether, these results confirm our

hypothesis that the effects of tDCS on motor performance require time. In the
 ACCEPTED MANUSCRIPT
17

analysis of the improvements during practice, we found that M1 stimulation was

associated with higher online gains. Although the first block of practice showed

no significant differences between groups, descriptively, G_M1 (206.2 ms)

T
presented a larger initial error than G_Sham (167.5 ms). This larger initial error

IP
may have given G_M1 more room for improvement.

R
In fact, the effects of tDCS on learning during practice should be more

SC
directly studied. For example, decreased performances and rates of learning

NU
were observed when anodal or cathodal tDCS were applied prior to the motor

task [27,33]. It has been speculated that positive feedback promoted by LTP-
MA
like plasticity destabilizes established cortical networks, leading to unregulated

cortical activity [27]. However, these studies included only the practice phase in
D

their experimental designs. If motor learning is observed through relatively

TE

permanent changes [24], the initial transitory effects of tDCS on behavior

P

cannot represent the steadier effects observed in the post-practice period. In

CE

this study, the stimulation prior to the practice phase allowed us to investigate

this gap in the literature. Overall, our findings indicated that the effects of M1
AC

stimulation were not observed in the performance during practice but were

detected in the post-practice period, specifically in the transfer test.

Our main finding is that M1 stimulation was associated with the absolute

dimension of the task. The advantage of practicing with only one skill during a

training session (called constant practice) on the relative dimension is well

known. Previous studies using this type of sequential task [2,23,34,35] have

demonstrated that the trial-to-trial stability produced by the skill repetition in

constant practice has a greater impact on the learning of the relative timing
 ACCEPTED MANUSCRIPT
18

dimension of the task than on the absolute dimension. Conversely, the best

practice schedule to learn the absolute dimension is variable random practice,

which is characterized by the non-systematic order of skills execution and a

T
non-consecutive execution of the same skill [23].

IP
During the practice phase, random practice is associated with increased

R
hemodynamic activity in the M1 throughout movement preparation and

SC
execution compared to more repetitive practice schedules [23,36,37]. Random

NU
practice also produces high levels of M1 excitability during skill practice, which

is positively associated with the level of skill consolidation observed in the

MA
retention test [38]. It should be noted that these studies did not separately

investigate the absolute and relative dimensions of the task. In the present
D

study, we expected the increased excitability of M1 to be associated with

TE

learning in the G_M1 group because tDCS elicits long-lasting, polarity-

P

dependent changes in cortex excitability [7]. The benefit observed in motor

CE

learning was identical to that found in behavioral studies using random practice,

that is, the learning of absolute dimension [23,35,39].

AC

Altogether, our results suggest that M1 has a functional contribution not

only to the production of length, direction, force and the time derivative of force

[6–8] but also to the temporal parameterization when the learner must learn

both dimensions of the task. Furthermore, our results indicate that M1 is

associated with the learning of the task when observed in the transfer test. In

the transfer test, the G_M1 group performed better than the G_SH group.

Again, the difference was in the absolute dimension.

 ACCEPTED MANUSCRIPT
19

Motor learning can be inferred not only by the ability to recall an

established motor skill [40] but also by the ability to adapt the motor commands

to new conditions and variants of the skill [41]. A transfer test is more difficult

T
than a retention test for those who practice using a constant schedule. In our

IP
study, motor practice without stimulation produced sufficient gain to later evoke

R
practiced skill at the same level of the motor practice with stimulation; however,

SC
when a variant of the skill practiced was required in the transfer test, the

stimulation of M1 produced better transfer of learning. The results related to the

NU
fast and slow learning indicate that the main contribution of the memory
MA
processes involving M1 was the long-term memory produced by the slow

learning, particularly in the transfer test.

D

Studies of motor adaptation have shown that the cerebellum plays an

TE

important role in both fast and slow learning [16]. Our findings also suggest an
P

association between slow learning and M1. Kantak et al. [42,43] observed that
CE

M1 has a central role in the retention and transfer tests during constant practice.

Unlike studies of motor adaptation, studies of motor sequence learning require

AC

a learner to acquire one or more motor skills. Additionally, throughout learning,

there is decreased participation of the cortico-cerebellar system, while the

cortico-striatal system participates more effectively during the final phase of skill

acquisition. When the skill is elicited, the cortico-striatal system is reactivated

[44]. An interesting question for future work concerning motor sequence

learning is whether the cerebellum has a more prominent role in the fast

process than in the slow processes during practice.

 ACCEPTED MANUSCRIPT
20

Several studies have failed to identify whether cathodal stimulation has

an effect on learning [19,26,45]. Stagg et al. [27] assumed that anodal and

cathodal tDCS modulate distinct neuronal populations. While GABA is

T
decreased with anodal tDCS within the stimulated M1, cathodal tDCS reduces

IP
glutamatergic neuronal activity [46]. This type of change in GABA activity is

R
associated with motor learning. In contrast, the change in the levels of

SC
glutamate have not been associated with motor learning [27]. In a prospective

future study, we are interested in the effects of cathodal stimulation and whether

NU
the cathodal stimulation of M1 negatively affects only the absolute dimension of
MA
the task.

A limitation of all studies that use non-invasive stimulation methods (e.g.,

D

TMS and tDCS) is focality. While some studies have used larger electrodes (35
TE

cm2) [26,27,33], we adopted a smaller electrode (25 cm2) like that used by
P

Cantarero et al. [18] and Reis et al. [19] in an attempt to position the electrode
CE

more accurately on M1. The use of a high-definition 4 × 1 ring configuration in

computational models [47] resulted in increased spatial specificity. The effects

AC

of tDCS can be also focalized by reducing the size of the stimulation electrode

on the target area and by increasing the size of the reference electrode [48].

Thus, future studies should use this type of tDCS setup.

Taken together, our results indicate that M1 is an essential area for

learning the absolute timing dimension of a motor sequence and that the slow

learning process is associated with M1. The motor consistency and the motor

adaptability are associated with the relative and absolute dimensions of the

task, respectively [1]. The M1 stimulation seems to be more associated with

 ACCEPTED MANUSCRIPT
21

improved motor adaptability, or the increased ability to cause small fluctuations

in the spatio-temporal pattern without losing its identity [2]. Secondary motor

areas have been associated with the motor consistency, or the spatio-temporal

T
pattern of movement sequencing [49]. For example, the supplementary motor

IP
area (SMA) is activated during the planning of sequential finger tapping [50].

R
Further studies could investigate whether the anodal stimulation of the

SC
secondary motor areas (pre-motor area and SMA) induces better learning of the

relative dimension of the task.

Acknowledgements NU
MA
This study was supported by INCT-MM, which is financed by the Brazilian

agencies for research development: Conselho Nacional de Desenvolvimento

D
TE

Científico e Tecnológico (CNPq), Coordenação de Aperfeiçoamento de Pessoal

de Nível Superior (CAPES) and Fundação de Amparo à Pesquisa do estado de

P

Minas Gerais (FAPEMIG) (FAPEMIG: CBB-APQ-00075-09/CNPq

CE

573646/2008-2). Support was also provided by the FAPEMIG and PRPq -

AC

UFMG incentive program for publication in indexed journals.

References

[1] D.J. Glencross, H.T.A. Whiring, B. Abernethy, Motor control, motor

learning and the acquisition of skill: historical trends and future directions,

Int. J. Sport Psychol. 25 (1994) 32–52.

[2] G.M. Lage, M.A.S. Alves, F.S. Oliveira, L.R. Palhares, H. Ugrinowitsch, R.

Benda, The combination of practice schedules: effects on relative and

absolute dimensions of the task, J. Hum. Mov. Stud. 52 (2007) 21–35.

 ACCEPTED MANUSCRIPT
22

[3] R.A. Schmidt, Motor Schema Theory after 27 Years: Reflections and

Implications for New Theory, Res. Quartely Exerc. Sport. 74 (2003) 366–

375.

T
[4] P. Nachev, C. Kennard, M. Husain, Functional role of the supplementary

IP
and pre-supplementary motor areas., Nat. Rev. Neurosci. 9 (2008) 856–

R
869.

SC
[5] A.P.E. Roland, Brain Activation, Wiley-Liss, New York:, 1993.

NU
[6] M.F.S. Rushworth, K.A. Hadland, T. Paus, P.K. Sipila, Role of the Human

Medial Frontal Cortex in Task Switching: A Combined fMRI and TMS

MA
Study, J Neurophysiol. 87 (2001) 2577–2592.

[7] H. Vollmann, V. Conde, S. Sewerin, M. Taubert, B. Sehm, O.W. Witte, et

D

al., Anodal transcranial direct current stimulation (tDCS) over

TE

supplementary motor area (SMA) but not pre-SMA promotes short-term

P

visuomotor learning, Brain Stimul. 6 (2013) 101–107.

CE

[8] R. Kawashima, P.E. Roland, B.T. O’Sullivan, Fields in human motor areas
AC

involved in preparation for reaching, actual reaching, and visuomotor

learning: a positron emission tomography study, J. Neurosci. 14 (1994)

3462–3474.

[9] C.-H.J. Lin, B.E. Fisher, A.D. Wu, Y.-A. Ko, L.-Y. Lee, C.J. Winstein,

Neural correlate of the contextual interference effect in motor learning: A

kinematic analysis, J. Mot. Behav. 41 (2009) 232–242.

[10] G. Prichard, C. Weiller, B. Fritsch, J. Reis, Effects of Different Electrical

Brain Stimulation Protocols on Subcomponents of Motor Skill Learning,

Brain Stimul. 7 (2014) 532–540.

 ACCEPTED MANUSCRIPT
23

[11] J. Reis, E. Robertson, J.W. Krakauer, J. Rothwell, L. Marshall, C. Gerloff,

et al., Consensus: “Can tDCS and TMS enhance motor learning and

memory formation?,” Brain Stimul. 1 (2008) 363–369.

T
[12] M. A Nitsche, W. Paulus, Excitability changes induced in the human

IP
motor cortex by weak transcranial direct current stimulation., J. Physiol.

R
527 (2000) 633–639.

SC
[13] G. Cantarero, B. Tang, R. O’Malley, R. Salas, P. Celnik, Motor learning

NU
interference is proportional to occlusion of LTP-like plasticity, J. Neurosci.

33 (2013) 4634–4641.
MA
[14] H.M. Schambra, M. Abe, D. A. Luckenbaugh, J. Reis, J.W. Krakauer, L.G.

Cohen, Probing for hemispheric specialization for motor skill learning: a

D

transcranial direct current stimulation study, J. Neurophysiol. 106 (2011)

TE

652–661.
P

[15] M.A. Smith, A. Ghazizadeh, R. Shadmehr, Interacting Adaptive

CE

Processes with Different Timescales Underlie Short-Term Motor Learning,

PLoS Biol. 4 (2006) 1035–1043.

AC

[16] K.M. Trewartha, A. Garcia, D.M. Wolpert, J.R. Flanagan, Fast But

Fleeting: Adaptive Motor Learning Processes Associated with Aging and

Cognitive Decline, J. Neurosci. 34 (2014) 13411–13421.

[17] E. Dayan, L.G. Cohen, Neuroplasticity subserving motor skill learning,

Neuron. 72 (2011) 443–454.

[18] G. Cantarero, A. Lloyd, P. Celnik, Reversal of long-term potentiation-like

plasticity processes after motor learning disrupts skill retention, J.

Neurosci. 33 (2013) 12862–12869.

 ACCEPTED MANUSCRIPT
24

[19] J. Reis, H.M. Schambra, L.G. Cohen, E.R. Buch, B. Fritsch, E. Zarahn, et

al., Noninvasive cortical stimulation enhances motor skill acquisition over

multiple days through an effect on consolidation, Proc. Natl. Acad. Sci. U.

T
S. A. 106 (2009) 1590–1595.

IP
[20] N.B. Albert, E.M. Robertson, R.C. Miall, The Resting Human Brain and

R
Motor Learning, Curr. Biol. 19 (2009) 1023–1027.

SC
[21] J.L. Mcgaugh, J.L. Mcgaugh, Memory — a Century of Consolidation,

NU
Science, 248 (2012) 248-251.

[22] E.M. Robertson, A. Pascual-Leone, R.C. Miall, Current concepts in

MA
procedural consolidation, Nat. Rev. Neurosci. 5 (2004) 576–582.

[23] G.M. Lage, H. Ugrinowitsch, T. Apolinário-Souza, M.M. Vieira, M.R.

D

Albuquerque, R.N. Benda, Repetition and variation in motor practice: A

TE

review of neural correlates, Neurosci. Biobehav. Rev. 57 (2015) 132–141.

P

[24] R.A. Schmidt, T.D. Lee, Motor control and learning: a behavioral
CE

emphasis, 4th ed., Human Kinetics, Champaign, IL, 2005.

AC

[25] R.D. Seidler, Neural correlates of motor learning, transfer of learning, and

learning to learn, Exerc. Sport Sci. Rev. 38 (2010) 3–9.

[26] M. A. Nitsche, A. Schauenburg, N. Lang, D. Liebetanz, C. Exner, W.

Paulus, et al., Facilitation of implicit motor learning by weak transcranial

direct current stimulation of the primary motor cortex in the human, J.

Cogn. Neurosci. 15 (2003) 619–26.

[27] C.J. Stagg, G. Jayaram, D. Pastor, Z.T. Kincses, P.M. Matthews, H.

Johansen-Berg, Polarity and timing-dependent effects of transcranial

direct current stimulation in explicit motor learning, Neuropsychologia. 49

 ACCEPTED MANUSCRIPT
25

(2011) 800–804.

[28] R.C. Oldfield, The assessment and analysis of handedness: the

Edinburgh inventory, Neuropsychologia. 9 (1971) 97–113.

T
[29] W. Paulus, Transcranial electrical stimulation (tES - tDCS; tRNS, tACS)

IP
methods, Neuropsychol. Rehabil. 21 (2011) 602–617.

R
[30] B. Fritsch, J. Reis, K. Martinowich, H.M. Schambra, Y. Ji, L.G. Cohen, et

SC
al., Direct current stimulation promotes BDNF-dependent synaptic

NU
plasticity: potential implications for motor learning, Neuron. 66 (2010)

198–204.
MA
[31] D.J. Kidgell, A.M. Goodwill, A.K. Frazer, R.M. Daly, Induction of cortical

plasticity and improved motor performance following unilateral and

D

bilateral transcranial direct current stimulation of the primary motor cortex,

TE

BMC Neurosci. 14 (2013) 64.

P

[32] Y.-J. Kwon, J.-K. Lee, D.-H. Shin, J.-S. Jeong, Changes in brain
CE

activation induced by the training of hypothesis generation skills: an fMRI

AC

study, Brain Cogn. 69 (2009) 391–397.

[33] M.-F. Kuo, M. Unger, D. Liebetanz, N. Lang, F. Tergau, W. Paulus, et al.,

Limited impact of homeostatic plasticity on motor learning in humans,

Neuropsychologia. 46 (2008) 2122–8.

[34] Q. Lai, C.H. Shea, G. Wulf, D.L. Wright, Optimizing generalized motor

program and parameter learning, Res. Q. Exerc. Sport. 71 (2000) 10–24.

[35] C.H. Shea, Q. Lai, D.L. Wright, M. Immink, C. Black, Consistent and

variable practice conditions: effects on relative and absolute timing, J Mot

Behav. 33 (2001) 52–139.

 ACCEPTED MANUSCRIPT
26

[36] E. Cross, P. Schmitt, S. Grafton, Neural substrates of contextual

interference during motor learning support a model of active preparation,

J. Cogn. Neurosci. 19 (2007) 1854–1871.

T
[37] N.N.F. Wymbs, S.T.S. Grafton, Neural substrates of practice structure

IP
that support future off-line learning, J. Neurophysiol. 102 (2009) 2462–

R
2476.

SC
[38] C.-H.J. Lin, B.J. Knowlton, M.-C. Chiang, M. Iacoboni, P. Udompholkul,

NU
A.D. Wu, Brain-behavior correlates of optimizing learning through

interleaved practice, Neuroimage. 56 (2011) 1758–1772.

MA
[39] H. Sekyia, R.A. Magill, D.I. Anderson, The contextual interference effect

in parameter modifications of the same generalized motor program, Res.

D

Q. Exerc. Sport. 67 (1996) 59–68.

TE

[40] R. Shadmehr, T. Brashers-krug, Functional stages in the formation of

P

human long-term motor memory, J. Neurosci. 17 (1997) 409–419.

CE

[41] R. Shadmehr, M. a Smith, J.W. Krakauer, Error correction, sensory

AC

prediction, and adaptation in motor control, Annu. Rev. Neurosci. 33

(2010) 89–108.

[42] S.S. Kantak, K.J. Sullivan, B.E. Fisher, B.J. Knowlton, C.J. Winstein,

Neural substrates of motor memory consolidation depend on practice

structure, Nat. Neurosci. 13 (2010) 923–925.

[43] S.S. Kantak, K.J.K. Sullivan, B.E. Fisher, B.J. Knowlton, C.J. Winstein,

Transfer of Motor Learning Engages Specific Neural Substrates During

Motor Memory Consolidation Dependent on the Practice Structure, J.

Mot. Behav. 43 (2011) 499–507.

 ACCEPTED MANUSCRIPT
27

[44] J. Doyon, V. Penhune, L.G. Ungerleider, Distinct contribution of the

cortico-striatal and cortico-cerebellar systems to motor skill learning,

Neuropsych.41 (2003) 252–262.

T
[45] J.M. Galea, P. Celnik, Brain polarization enhances the formation and

IP
retention of motor memories., J. Neurophysiol. 102 (2009) 294–301.

R
[46] C.J. Stagg, J. O’Shea, Z.T. Kincses, M. Woolrich, P.M. Matthews, H.

SC
Johansen-Berg, Modulation of movement-associated cortical activation by

NU
transcranial direct current stimulation., Eur. J. Neurosci. 30 (2009) 1412–

23.
MA
[47] M. Bikson, A. Rahman, A. Datta, Computational Models of Transcranial

Direct Current Stimulation, Clin. EEG Neurosci. 43 (2012) 176–183.

D

[48] M.A.Nitsche, S. Doemkes, T. Karako¨se, A. Antal, D. Liebetanz, N. Lang, F.

TE

Tergau, W. Paulus, Shaping the effects of transcranial direct current

P

stimulation of the human motor cortex., J Neurophysiol. 97 (2007) 3109-

CE

3117.
AC

[49] P.E. Roland, B. Larsen, N. A Lassen, E. Skinhøj, Supplementary motor

area and other cortical areas in organization of voluntary movements in

man., J. Neurophysiol. 43 (1980) 118–36.

[50] G.H. Chung, Y.M. Han, S.H. Jeong, C.R. Jack, Functional heterogeneity

of the supplementary motor area, AJNR. Am. J. Neuroradiol. 26 (2005)

1819–1823.
 ACCEPTED MANUSCRIPT
28

FIGURE CAPTIONS

Fig. 1 - Motor consistency (represented in red) and motor adaptability (in blue)

in a sequential motor task.

T
IP
Fig. 2 - Experimental settings. A = anodal electrode over right C3; C = cathodal

R
electrode over right supraorbital.

SC
Fig. 3 - Apparatus used to apply the motor task, the sequence of keys typed

NU
(K2…K8), the relative criterion segment ratios between keys (22.2%, 44.4%,

and 33.3%) and the total criterion movement time (900 ms).
MA
Fig. 4 – The model multi-rate learning process (Smith et al., 2006) used to

calculate fast and slow learning in the absolute dimension. Compared to the
D

slow process, the fast process presented a higher learning rate (βf > βs) but a
TE

low retention factor (αf < αs). The motor output is the combination of both
P

outputs of the memory process. The performance error is the difference

CE

between the motor output and the MT criterion.

AC

Fig. 5 - Absolute dimension analyses of G_M1 (in orange) and G_SH (in yellow)

groups. (a) Absolute timing error (AE) during practice (blocks 1 to 10) and the

retention (Rt) and transfer (Tt) tests. (b) Improvements during practice (online

effects) and post-practice (offline effects in Rt and Tt). (c) Levels of fast and

slow learning in the Rt and Tt tests. (d) Intersubject variability over the blocks of

trials during practice (bl1…bl10) and the Rt and Tt tests.

Fig. 6 - Relative dimension analyses of G_M1 (orange) and G_SH (yellow)

groups. (a) Absolute timing error (AE) during practice (blocks 1 to 10) and
 ACCEPTED MANUSCRIPT
29

retention (Rt) and transfer (Tt) tests. (b) Improvements during practice (online

effects) and post-practice (offline effects in Rt and Tt). (c) Relative timing

among movement components (c1, c2, c3). (d) Inter-subject variability over the

T
blocks of trials during practice (bl1…bl10) and Rt and Tt tests.

R IP
SC
NU
MA
D
P TE
CE
AC
 ACCEPTED MANUSCRIPT
30

T
RIP
SC
NU
MA
D
TE
P

Figure 1
CE
AC
 ACCEPTED MANUSCRIPT
31

T
RIP
SC
NU
MA
D

Figure 2
TE
P
CE
AC
 ACCEPTED MANUSCRIPT
32

T
RIP
SC
NU
MA
D
TE
P
CE

Figure 3
AC
 ACCEPTED MANUSCRIPT
33

T
RIP
SC
NU
MA
Figure 4
D
TE
P
CE
AC
 ACCEPTED MANUSCRIPT
34

T
RIP
SC
NU
MA
D

Figure 5
TE
P
CE
AC
 ACCEPTED MANUSCRIPT
35

T
RIP
SC
NU
MA
D

Figure 6
TE
P
CE
AC
 ACCEPTED MANUSCRIPT
36

Highlights

 Consistency and adaptability are characteristics of the skilled behavior.

 Primary motor cortex (M1) may be more involved in the learning of the

T
adaptability.

IP
 We stimulate M1 to facilitate motor learning.

R
 The stimulation of M1 affected more the acquisition of motor adaptability.

SC
NU
MA
D
P TE
CE
AC