Neuropsychologia ∎ (∎∎∎∎) ∎∎∎–∎∎∎

Contents lists available at ScienceDirect

Neuropsychologia journal homepage: www.elsevier.com/locate/neuropsychologia

Neurophysiological correlates of visuo-motor learning through mental and physical practice Nadia Allami a,1, Andrea Brovelli b,1, El Mehdi Hamzaoui c, Fakhita Regragui c, Yves Paulignan a, Driss Boussaoud d,n a

Institut des Sciences Cognitives, UMR5230, 67, Bd Pinel—69675 Bron Cedex, France Institut de Neurosciences de la Timone (INT), UMR 7289 CNRS, Aix Marseille University, Campus de Santé Timone, 27 Bd. Jean Moulin, 13385 Marseille, France c Laboratoire d0 Informatique, Mathematiques Appliquées, Intelligence Artificielle et Reconnaissance de Formes—LIMIARF, Université Mohammed V—Agdal, Faculté des Sciences, Rabat, Maroc d Institut de Neurosciences des Systèmes (INS), UMR 1106 INSERM, Aix Marseille University, Campus de Santé Timone, 27 Bd. Jean Moulin, 13385 Marseille, France b

art ic l e i nf o

Keywords: Motor learning Neuroplasticity Mental imagery EEG

a b s t r a c t We have previously shown that mental rehearsal can replace up to 75% of physical practice for learning a visuomotor task (Allami, Paulignan, Brovelli, & Boussaoud, (2008). Experimental Brain Research, 184, 105– 113). Presumably, mental rehearsal must induce brain changes that facilitate motor learning. We tested this hypothesis by recording scalp electroencephalographic activity (EEG) in two groups of subjects. In one group, subjects executed a reach to grasp task for 240 trials. In the second group, subjects learned the task through a combination of mental rehearsal for the initial 180 trials followed by the execution of 60 trials. Thus, one group physically executed the task for 240 trials, the other only for 60 trials. Amplitudes and latencies of event-related potentials (ERPs) were compared across groups at different stages during learning. We found that ERP activity increases dramatically with training and reaches the same amplitude over the premotor regions in the two groups, despite large differences in physically executed trials. These findings suggest that during mental rehearsal, neuronal changes occur in the motor networks that make physical practice after mental rehearsal more effective in configuring functional networks for skilful behaviour. & 2013 Elsevier Ltd. All rights reserved.

1. Introduction Thanks to the important contribution of Jeannerod et al. (Decety et al., 1989; Frak et al., 2001; Jeannerod & Frak, 1999), it is nowadays accepted that mentally represented and physically executed actions are mediated by overlapping cortical and sub-cortical networks. This idea gained support from neuroimaging studies showing that motor imagery activates the motor system, including cortical areas such as the premotor cortex (Binkofski et al., 2000; Gerardin et al., 2000; Naito et al., 2002; Rizzolatti et al., 1996), the supplementary motor area (SMA; Lotze et al., 1999; Naito et al., 2002; Seitz et al., 2000; Solodkin et al., 2004), parietal cortex (Gerardin et al., 2000; Nair et al., 2003; Sirigu et al., 1996; Wolbers et al., 2003) and even the primary motor area (M1; Ehrsson et al., 2003; Nair et al., 2003; Porro et al., 1996; Roth et al., 1996; Sabbah et al., 1995). Although less numerous, electroencephalographic (EEG) studies also contributed to

n

Corresponding author. Tel.: þ 33 62 2263752; fax: þ33 49 1789914. 1 Authors contributed equally.

support the view of equivalence between mentally and physically executed action (Beisteiner et al., 1995; Caldara et al., 2004; Cunnington et al., 1996; Kilner et al., 2004; Naito & Matsumura, 1994; Romero et al., 2000). Taken together, these results have led to the hypothesis that motor imagery (similarly to physical execution) may induce motor learning, and opened the way to the use of motor imagery as a method to improve motor performance (Jackson et al., 2001). On the one hand, mental practice improves motor performance (Allami et al., 2008; Debarnot et al., 2011; Gentili et al., 2010; Gentili et al., 2006). On the other hand, several studies have reported that improved motor performance after physical practice correlates with changes in functional brain networks including motor and premotor cortex, cerebellum, basal ganglia and the fronto-parietal networks (Doyon et al., 2003; Karni et al., 1995; Lacourse et al., 2004; Nyberg et al., 2006; Pascual-Leone et al., 1995). Finally, some studies have compared these functional changes following physical and mental practice and reported similar brain plasticity following both types of practice (Debarnot et al., 2011; Jackson et al., 2003; Lacourse et al., 2004; Lafleur et al., 2002; Nyberg et al., 2006).

0028-3932/$ - see front matter & 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.neuropsychologia.2013.12.017

Please cite this article as: Allami, N., et al. Neurophysiological correlates of visuo-motor learning through mental and physical practice. Neuropsychologia (2014), http://dx.doi.org/10.1016/j.neuropsychologia.2013.12.017i

N. Allami et al. / Neuropsychologia ∎ (∎∎∎∎) ∎∎∎–∎∎∎

2

However, a direct neurophysiological comparison between the neuronal changes occurring after learning through physical or mental practice is still lacking. To our knowledge, only one electrophysiological study has investigated the changes in event-related potentials (ERPs) in relation with physical practice alone (Staines et al., 2002), but no study has compared physical and mental practice directly using electrophysiological technique, yet. In a previous report (Allami et al.,2008), we have shown that healthy subjects learn to perform a reach-to-grasp task at the same rate, or even faster, if 75% of physical practice is replaced by mental rehearsal, as compared to 100% physical practice. Here, we sought to investigate the neural underpinnings of such equivalence between imagined and executed actions in the context of motor learning. To do so, we used EEG recordings to compare the ERPs measured during the practice of a reach-to-grasp task, through physical practice alone or through a combination of mental and physical practice. The goal of the current study was to exploit previously identified optimal parameters (Allami et al., 2008) to test the hypothesis that during mental practice, functional changes take place in the motor networks that lead to a gain in behavioural performance. We have targeted an ERP negative component (N2), which occurs approximately 300 ms after the go signal, just before the subjects start to move their arm. Previous studies have shown that the N2 is observed over cortical networks involved in motor preparation and sensorimotor integration, including the parietofrontal and frontal regions (Falkenstein et al., 1999; Romero et al., 2000; Staines et al., 2002). More importantly, the N2 component was shown to reflect ERP changes during physical practice (Staines et al., 2002) over the parieto-frontal and premotor regions. The data reported in this paper show that ERP changes occur in these brain regions following both physical and mental practice. We suggest that during mental rehearsal, sensorimotor and premotor regions undergo neuronal changes that make physical practice more effective in inducing ERP changes that mediate the improvement in behavioural performance.

Marble

56° Object

S1

Support

0°

S2

Ready

Open Eyes

Execution or Mental imagery Close Eyes

0.28 3.3 s

FP1

FPz

FP2

AFz F7 F3 FC3 FC1 T3

C3

C1

P3 O1

F8

FC2 FC4 Cz

CP3 CP1 P7

F4

Fz

C4

C2

T4

CP2 CP4 Pz Oz

P4

P8

O2

2. Methods 2.1. Subjects and behavioural paradigm Ten right-handed subjects aged 21 to 34 years (average age 27.5 years) were volunteers in the study. All were naïve as to the purpose of the experiment and did not have explicit knowledge concerning motor imagery processes. None of the participants had history of nervous or muscular disorders, and they all gave written informed consent according to established institutional guidelines. The project has been approved by the local ethics committee. Subjects were randomly included in one of two groups: the first group of subjects executed physically a total of 240 trials (this will be referred to as execution group, GEx). The second group first rehearsed the task mentally in 75% of 240 trials (180 trials), and then executed physically the remaining 25% of the trials (60 trials). This group will be referred to as the “imagination group” (GIm). The task has been previously described in detail (Allami et al., 2008). Briefly, we used a two-step task, where subjects reached for and grasped an object, and transported it to insert it in an adapted support (Fig. 1). The subjects were comfortably seated in an adjustable chair in front of a table with the right hand resting palm down on the starting point located 29 cm to the right side of the sagittal axis. Subjects were asked to grasp a plastic parallelepiped (first movement) and insert it in a support. The object was located along the subject0 s sagittal axis, 38 cm from the chest. Half of the object surface was coloured in grey, the other half was white with black marks that matched in location with identical marks on the support. Subjects had to grasp the object from its grey side and to place it carefully inside the support. To make the execution of the task more difficult, a marble was inserted in an unstable manner in a slight hole made on the object0 s surface. Furthermore, two small wooden sticks were glued on the object0 s small sides to force the subjects to grasp with a precision grip. Before the learning session, all subjects physically executed 5 trials to familiarize with the set-up. During the familiarisation period, subjects practised how to position their fingers on the object and how to insert it in the support. This allowed them to assess the instability of the marble.

Fig. 1. Setup, behavioural task and EEG recordings.(A) Illustration of the experimental setup. The subjects sits with her right arm on the table, and is asked to left a parallelepiped object and insert into a slot made in another distant object (support). The trial is composed of two movements: the 1st movement consisted in reaching to and grasping the object with a precise grip (thumb and index only); the second movement consisted in taking the object to insert it into the support. The orientation of the support (01) was the same throughout the trials, but the orientation of the object changed from trial to trial, in pseudo-random way (221, 01, 451 and 561). To make the execution of the task more difficult, a marble was inserted in a slight hole made on the object0 s surface. Furthermore, two small wooden sticks were glued on the object0 s small sides to force the subjects to grasp with a precision grip. The bottom part shows the sequence of events. Subject opens eyes at the 1st sound (S1), and waits for another sound (S2) which serves as the gosignal. Then, the subject either executes the trial, or rehearses the execution of movement, then closes eyes and waits for S1. (B) EEG recordings. Electrode location on the scalp (Brain Vision Recorder), positioned according to the international modified 10–20 system.

2.1.1. Physical practice The sequence of the task is illustrated at the bottom of Fig. 1. Before each trial, subjects were told to close their eyes, allowing the experimenter to place the object on the table and vary its orientation pseudo-randomly from trial to trial. A first acoustic tone (S1) instructed the subjects to open their eyes and wait for a second tone (S2), which instructed them to initiate the movement (the go signal). The gosignal occurred after a variable delay ranging between 0.28 and 3.3 s, and informed the subjects to reach for and grasp the object with a precision grip (between thumb and index), transport it to the support and insert it correctly in the slot (i.e. the marks on the object must match those of the support; Fig. 1). At the end of the trial, subjects returned their hand to the starting position and closed their eyes, waiting for the next trial to start. No feedback was given to the subjects about their performance.

Please cite this article as: Allami, N., et al. Neurophysiological correlates of visuo-motor learning through mental and physical practice. Neuropsychologia (2014), http://dx.doi.org/10.1016/j.neuropsychologia.2013.12.017i

N. Allami et al. / Neuropsychologia ∎ (∎∎∎∎) ∎∎∎–∎∎∎ 2.1.2. Mental practice The subjects were instructed to follow the same sequence of events as in the physical practice, except that they were told to imagine and feel themselves (first person) doing the task as if real (i.e. grasp the object, lift it and transport it to the support). When they completed the “mental” trial, they had to press a switch, close their eyes and wait for the next trial. The reaction (RT) and movement (MT) times for the physically executed actions, and the end of the mentally rehearsed actions were recorded using a switch positioned under the subjects0 hand. The RT corresponds to the time between the go signal and hand lifting to reach for the object, whereas the MT corresponds to the time between hand lifting and its return after completion of the movement. 2.2. EEG recordings The EEG activity was recorded using a 32 electrodes system (Brain Vision Recorder), positioned according to a modified international 10–20 system that covered more densely the sensorimotor and premotor areas (Fig. 2). The horizontal and vertical electro-oculographic (EOGh and EOGv) movements were also recorded. EEG data was recorded at a sampling frequency of 1 kHz and stored for off-line analysis. Raw EEG data were band-pass filtered between 2 and 15 Hz, and average-referenced. Channels with large contaminations from eye movements and/ or muscular artefacts were excluded from the subsequent analysis. Fifteen electrodes were found to be artefact free across subjects, among which were the sites of interest for this study, i.e. the fronto-central and central sites (FC1, FC3, C1 and C3 on the left hemisphere, contralateral to the hand used; FC2, FC4, C2 and C4 on the right hemisphere). The EEG epochs aligned to the go signal (S2) were created and event-related potentials (ERPs) were computed for each subject from 40 artefactfree trials taken at the beginning, middle and end of the session for each group. Then, the grand average across all subjects was computed for the execution group (GEx) and for the imagination group (GIm) for the three learning levels. 2.3. Statistical analysis of behavioural and EEG data Statistical analysis of RTs and MTs was performed using two-sample t-test. For each group, the RTs and MTs of the early and late learning phases were compared and the statistical significance was assessed. To quantify differences across groups, we compared the RTs and MTs across groups in the last phase of training. The same number of trials used for the EEG analysis was used for behavioural analysis. To assess statistical differences in mean amplitudes and latencies of the N2 peak across learning levels and experimental groups, we used a two-way ANOVA. Learning level was set as the first factor (three levels: beginning, middle, end of learning) and group type as the second factor (two levels: GEx and Gim). Post-hoc analyses (taking into account multiple comparisons) were conducted using the Newman–Keuls method and a po 0.05 was chosen as the level of significance.

3. Results 3.1. Behavioural performance Behavioural data indicate that subjects learned in both experimental groups, i.e. their performance improved between the

400

a

5000

b

4000

300

3000

200

2000

100

1000

0

Movement Time (ms)

Reaction Time (ms)

500

0 GEx early

GEx end

GIm end

Fig. 2. Comparison of reaction times (RTs, left axis) and movement times (MTs, right axis) across groups. The bars in dotted lines represent RTs, those in solid lines represent MTs for early phase of learning in group GEx (GEx early), and late phase of learning in group GEx and GIm, respectively. Significant differences between GEx early and the other two phases are indicated by letters (a) for RTs, (b) for MTs (p o 0.05).

3

beginning and the end of the training session (Fig. 2). In the GEx group, the mean RT (indicated as (a) in Fig. 2) decreased from 472 722 ms (mean7 standard deviation) early in the training session to 380 ms ( 725 ms) at the end. The results of a twosample t-test showed that the difference was statistically significant (t(8)¼2.795, p ¼0.02). In the GIm group, the RT at the end of the session was 384 ms (729 ms), and differed significantly from the mean RT of GEx at the beginning of session, suggesting that mental practice led to improvement in performance (t(8)¼ 2.44, p¼ 0.0402). The mean RT did not differ significantly between the two groups at the end of the learning session (t(8) ¼  0.101, p¼ 0.92). Analysis of movement times (MT, indicated as (b) in Fig. 2) showed that the mean MT in GEx decreased significantly at the late phase of learning (2952 7133 ms), relative to the early phase (34907 105 ms). After learning by mental practice, followed by physical practice (GIm), the mean MT reached 3050 7178 ms, thus showing a significant difference (t(8) ¼2.036, p ¼0.038) with respect to the early phase of learning in GEx, but did not differ (t (8)¼  0.019, p ¼0.507) from mean MT at the end of learning in GEx. These results replicate the behavioural findings previously reported (Allami et al., 2008) and demonstrate improvement in performance following combination of mental and physical practice. 3.2. ERPs activity: General pattern The general ERP topography, at the end of the training session shows a prominent left lateralised negative activity before movement onset, over the fronto-central and central sites in both groups (Fig. 3). This negative activity occurs approximately 300 ms after the go-signal (i.e. during the reaction time period) and corresponds to the N2 component, which was previously shown to change during physical practice (Staines et al., 2002). We will focus our analysis on this component in order to determine the changes in its amplitude and latency during training in both experimental groups. Early during the training session, whether through physical or mental practice, ERPs are generally weak. Fig. 4 shows examples of these ERPs for four key electrodes (FC1, FC2, C1 and C2), computed by averaging the 40 first executed in Gex (top panel) or imagined GIm trials (bottom panel), corresponding to the beginning of learning. They are characterized by a succession of low amplitude waves: a first negative component (N1) which peaks at 100– 120 ms, followed by a positive component (P2) whose peak is at 168–220 ms, then a second negative component (N2) at 265– 302 ms. Even though the waveforms are similar across the two groups (GEx and GIm), two-way ANOVA on the amplitude and latency of the N2 peak, with learning level and group type as factors, revealed important findings (Table 1). Comparison across groups showed that the N2 amplitudes are not significantly different (no group effect in the top panel of Table 1). However, a significant difference across learning levels was present (p ¼0.0028). Similarly, for what concerns the analysis of N2 latencies, the only significant effect was found for the learning factor (bottom panel of Table 1). 3.3. Learning with physical practice only (GEx) To further investigate the ERP correlates of learning through physical practice alone (GEx), the ERPs computed for each subject at the beginning of the training session (mean of the first 40 trials) were compared with those recorded at the end of the session (averaged over the 40 last trials). This comparison indicates that the amplitude of the N2 component increases dramatically during learning (Fig. 5). Interestingly, this strong increase occurs in the fronto-central sites (FC1, FC2, FC3), in sharp

Please cite this article as: Allami, N., et al. Neurophysiological correlates of visuo-motor learning through mental and physical practice. Neuropsychologia (2014), http://dx.doi.org/10.1016/j.neuropsychologia.2013.12.017i

N. Allami et al. / Neuropsychologia ∎ (∎∎∎∎) ∎∎∎–∎∎∎

4

2.5

FC3

2.5

FC1

2.5

2.5

FC2

0

0

0

0

-2.5

-2.5

-2.5

-2.5

2.5

2.5

C3

0 -2.5 2.5

CP3

C1

2.5

C2

2.5

0

0

0

-2.5 2.5

-2.5 2.5

-2.5

CP1

CP2

2.5

0

0

0

-2.5

-2.5 2.5 2.5 P3

-2.5 Pz

-2.5

Amplitude(µV)

0

00

-2.5 -2.5

0

2.5

2.5

0

0

-2.5 500 Time (ms)

-2.5

FC4

C4

CP4

*

P4

Fig. 3. General topography of ERP activity. (A). Grand average ERPs for 15 electrodes shown at their respective location on the head, seen from above. Front is up. (B) and (C) represent colour coded maps of the ERPs activity at the end of learning in groups GEx and GIm, respectively. Note the negative activity over the fronto-central electrodes, measured between 280 and 330 ms after the go signal.

contrast with the central sites (C1, C2, C3) where the changes between the early and the last phases of learning remain modest. No clear difference seemed to be present between the N2 latencies in the early and the late phases of learning in the GEx group (Fig. 5). This suggest that may learning effect on N2 latencies shown in Table 1 (bottom panel) arised from the comparison with the intermediate group. In fact, Fig. 7 (top panel) shows that the main difference in latency of the N2 components arises from

the comparison between the intermediate and the late stages of learning. 3.4. Learning with combination of mental and physical practice (GIm) As for group GEx, we compared the ERPs averaged across the first 40 trials rehearsed mentally, with those averaged across

Please cite this article as: Allami, N., et al. Neurophysiological correlates of visuo-motor learning through mental and physical practice. Neuropsychologia (2014), http://dx.doi.org/10.1016/j.neuropsychologia.2013.12.017i

N. Allami et al. / Neuropsychologia ∎ (∎∎∎∎) ∎∎∎–∎∎∎

5

Finally, we compared the ERPs across the two groups (GEx and GIm), and found no difference neither in the amplitude nor in the latencies of the N2 component. Fig. 6 shows the average ERPs recorded in the two groups for each of the fronto-central and central sites. This figure depicts the main finding summarised in Table 1, displaying no group effect neither in the amplitude nor in the latency of the N2. 3.5. Dynamics of ERPs during learning

Amplitude(µV)

N1

FC1 FC2 C1 C2

Fig. 7 compares the average ERP waveforms across the two groups, for three learning stages: the first, intermediate and last 40 trials. We performed a two-way ANOVA to dissociate the effects induced by learning stage (3 levels: early, intermediate and late) and learning type (GEx, GIm). The main effect of learning stage was found to be significant on N2 amplitude (F(2,24)¼7.612, po0.027). Comparison of the averaged activity (Newman–Keuls post-hoc test) for N2 component between the middle and the beginning of learning did not show significant differences (p¼0.25), in contrast to the comparison with the end of learning where significant differences were observed (p¼0.002). At the end of learning, no group effect was observed: the amplitudes of the N2 component were not statistically different across groups (F(1,24)¼ 0.735; p¼0.4), and there were no interaction effect between the two factors, i.e. learning level and group (F(2,24)¼0.045, p¼0.9).

4. Discussion

100

200

300

400

500

Time (ms) Fig. 4. Mean ERP activity at the beginning of learning. ERPs at four fronto-central and central sites are averaged for the first 40 executed trials in GEx (top), and first 40 imagined trials in GIm (bottom).

Table 1 Statistical analyses of group and learning level effects on N2 amplitude and latency. Group and learning level—N2 amplitude Source

df

Group 1 Learning level 2 Interaction 2 Group and learning level—N2 latency Group 1 Learning level 2 Interaction 2

F

p Value

0.74 7.61 0.04

0.399 0.0028 0.956

0.51 4.03 0.59

0.48 0.031 0.564

the last 40 trials (executed physically). Similarly to the GEx group, the ERPs recorded in the GIm group showed major increases in the amplitude of the N2 component over the fronto-central sites (FC1, FC2 and FC3) at the end of the session (Fig. 5, bottom). Interestingly, the changes of ERPs over the central sites (C1 and C2) were larger and the differences between early and late phases reached significance for the C1 electrode. Contrary to the GEx group, the differences in N2 latencies between early and late stages of learning was significant, the latencies being shorter at the late phase of learning (as compared to early phase) at one frontocentral site (FC1) and one central site (C1).

From the behavioural perspective, mental rehearsal is known to help athletes improve their skills (Driskell et al., 1994; Feltz & Landers, 1983; Janssen & Sheikh, 1994) and to facilitate motor recovery of both upper and lower limbs in stroke patients (De Vries & Mulder, 2007; Dijkerman et al., 2004; Jackson et al., 2004; Jackson et al., 2001; Page et al., 2001; Stevens & Stoykov, 2003; Weiss et al., 1994; Yoo et al., 2001). Laboratory experiments have provided a systematic analysis of the effects of mental rehearsal in healthy subjects (Allami et al., 2008; Gentili et al., 2010; Gentili et al., 2006; Mulder et al., 2004; Yaguez et al., 1998). What lacked was the characterisation of the underlying brain plasticity that allows mental rehearsal to improve motor learning. Previous neuroimaging studies have provided evidence that, during mental and physical practice, functional brain changes occur in similar and/or overlapping brain networks, including the cerebellum, basal ganglia, frontal and parietal cortex (Jackson et al., 2003; Lacourse et al., 2004; Lafleur et al., 2002; Nyberg et al., 2006). Here, we exploited the high temporal resolution of the EEG to directly compare the changes in well described ERP component (i.e. the N2 component) during learning through physical practice alone with those that occur during mental rehearsal and the subsequent physical practice. 4.1. Effects of mental rehearsal on learning-related neural plasticity Obviously, learning a motor task is not possible without some physical exercise, but mental rehearsal helps improve motor performance, and can replace a large amount of physical exercise (Allami et al., 2008). The question is whether mental rehearsal induces neuronal changes comparable to those observed during physical exercise, thereby accelerating practice-related changes in the motor networks. The present data show that mental practice does not induce significant changes of N2 component (same amplitude at the beginning and middle of the training session), nor does physical practice at the early and middle stages of learning. However, mental practice seems to “prime” ERP changes induced by the subsequent physical practice. Consequently, at the

Please cite this article as: Allami, N., et al. Neurophysiological correlates of visuo-motor learning through mental and physical practice. Neuropsychologia (2014), http://dx.doi.org/10.1016/j.neuropsychologia.2013.12.017i

N. Allami et al. / Neuropsychologia ∎ (∎∎∎∎) ∎∎∎–∎∎∎

6

3 0 0.2 0.4 0.6 -3

0.8

1

Time (s)

Significant difference in amplitude (p