Neuroscience 297 (2015) 231–242

LATERALITY EFFECTS IN MOTOR LEARNING BY MENTAL PRACTICE IN RIGHT-HANDERS R. J. GENTILI a,b,c* AND C. PAPAXANTHIS d,e

internal models theory. Ó 2015 IBRO. Published by Elsevier Ltd. All rights reserved.

a

Department of Kinesiology, School of Public Health, University of Maryland, College Park, MD, USA b

Neuroscience and Cognitive Science Graduate Program, University of Maryland, College Park, MD, USA

Key words: motor imagery, forward model, mental practice, internal model, arm reaching movement.

c Maryland Robotics Center, University of Maryland, College Park, MD, USA d

Universite´ de Bourgogne, Unite´ de Formation et de Recherche (UFR) en Sciences et Techniques des Activite´s Physiques et Sportives (STAPS), Dijon, France

INTRODUCTION

e

Institut National de la Sante´ et de la Recherche Me´dicale (INSERM), UMR 1093, Cognition, Action et Plasticite´ Sensorimotrice (CAPS), Dijon, France

During motor imagery practice subjects internally simulate a movement without any motor output. This mental process implies that individuals feel themselves performing a movement in a first-person perspective (e.g., imagined the sensation of shooting a basketball). Neurophysiological and psychophysical studies have revealed that mental and actual states of action trigger similar motor representations and share overlapping neural substrates (Jeannerod, 2001; Guillot and Collet, 2005; Lorey et al., 2009; Munzert et al., 2009). For instance, common activations of the parietal and prefrontal cortices, the supplementary motor area, the premotor and primary motor cortices, the basal ganglia, and the cerebellum have been repeatedly reported. Furthermore, the activation of the autonomic nervous system, such as heart and respiration rate, increases proportionally to the mental effort produced by subjects during mental movements (Decety et al., 1993; Demougeot et al., 2009; Collet et al., 2013). Lastly, mental actions preserve the same spatiotemporal characteristics and obey the same motor rules as their overt counterparts (Decety and Jeannerod, 1995; Papaxanthis et al., 2002, 2012; Gentili et al., 2004; Bakker et al., 2007). Mental training, by means of motor-imagery, is a potential tool in sports and motor rehabilitation, because it was shown to improve motor function. In particular, mental training enhances muscular force (Yue and Cole, 1992; Zijdewind et al., 2003; Ranganathan et al., 2004) and improves movement kinematics (Ya´gu¨ez et al., 1998; Gentili et al., 2006, 2010; Allami et al., 2008; Avanzino et al., 2009). The concept of internal model provides the theoretical basis to understand the positive effects of mental training on motor performance (e.g., Wolpert and Miall, 1996; Gentili et al., 2010). Forward internal models mimic the causal flow of the physical process, that is the mapping from motor commands to sensorimotor consequences, by predicting the future sensorimotor state (e.g., position, velocity) given the efferent copy of the motor command and the current state

Abstract—Converging evidences suggest that mental movement simulation and actual movement production share similar neurocognitive and learning processes. Although a large body of data is available in the literature regarding mental states involving the dominant arm, examinations for the nondominant arm are sparse. Does mental training, through motor-imagery practice, with the dominant arm or the nondominant arm is equally efficient for motor learning? In the current study, we investigated laterality effects in motor learning by motor-imagery practice. Four groups of right-hander adults mentally and physically performed as fast and accurately as possible (speed/accuracy trade-off paradigm) successive reaching movements with their dominant or nondominant arm (physical-training-dominant-arm, mental-training-dominant-arm, physical-training-nondominant-arm, and mental-training-nondominant-arm groups). Movement time was recorded and analyzed before, during, and after the training sessions. We found that physical and mental practice had a positive effect on the motor performance (i.e., decrease in movement time) of both arms through similar learning process (i.e., similar exponential learning curves). However, movement time reduction in the posttest session was significantly higher after physical practice than motor-imagery practice for both arms. More importantly, motor-imagery practice with the dominant arm resulted in larger and more robust improvements in movement speed compared to motor-imagery practice with the nondominant arm. No such improvements were observed in the control group. Our results suggest a superiority of the dominant arm in motor learning by mental practice. We discussed these findings from the perspective of the *Correspondence to: R. J. Gentili, Department of Kinesiology, 2144 School of Public School Building (#255), University of Maryland, College Park, MD 20742, USA. Tel: +1-301-405-2490; fax: +1-301405-5578. E-mail address: [email protected] (R. J. Gentili). Abbreviations: ANOVA, analysis of variance; RMS, root mean square; rMSE, root mean square of the error. http://dx.doi.org/10.1016/j.neuroscience.2015.02.055 0306-4522/Ó 2015 IBRO. Published by Elsevier Ltd. All rights reserved. 231

232

R. J. Gentili, C. Papaxanthis / Neuroscience 297 (2015) 231–242

(Wolpert and Miall, 1996). During physical training, the estimated state of the motor system can be employed to refine future motor commands by generating an internal training signal that modifies plastic neural processes (Wolpert et al., 1995; Kawato, 1999; Desmurget and Grafton, 2000). In addition, noisy and delayed sensory feedback is thought to be combined with forward model output to provide accurate and precise state estimation (Wolpert et al., 1995). During mental training, a similar plastic neural mechanism based on the estimated state of the motor system can be utilized (Wolpert and Miall, 1996; Gentili et al., 2006, 2010). Motor learning by mental training is associated with changes in brain activation both in healthy individuals (Lafleur et al., 2002; Jackson et al., 2003) and stroke patients (Page et al., 2009). However, since during mental training the state estimation derives from the forward model alone, the training signal is presumably less accurate and less precise than during physical training. This may explain in part why mental training is, in general, less efficient than physical training (Gentili et al., 2006, 2010). Though a large body of work regarding mental states related to the dominant arm performance is available, research for the nondominant arm is still relatively limited. Neurophysiological and clinical investigations (Fadiga et al., 1999; Lotze et al., 1999; Sabate´ et al., 2004; Stinear et al., 2006a), as well as psychophysical studies examining the temporal aspects of imagined arm movements (Maruff et al., 1999; Skoura et al., 2008, 2009), have shown that lateralization also emerges in mental imagery. An intriguing question, however, is whether mental training with the dominant or nondominant arm is equally efficient for motor learning. Previous studies have suggested that the left hemisphere/right arm control system would be predominantly involved in movement organization and selection (Haaland and Harrington, 1996; Schluter et al., 1998; Rushworth et al., 2001), in movement representation and learning (Grafton et al., 2002; Kuhtz-Buschbeck et al., 2003), and in body state estimation (Wolpert et al., 1998; Mutha et al., 2011), suggesting its important role in feedforward control processes (Sainburg, 2002; Agnew et al., 2004; Mutha et al., 2011, 2012). On the other hand, the right hemisphere/left arm control system appears to have reduced higher order planning functions (Amunts et al., 1996; Serrien et al., 2006), with preferential involvement in feedback control processes (Sainburg, 2002; Mutha et al., 2012). According to this potential mechanism, (i.e., the dynamic-dominant hypothesis) one could expect better learning by mental practice for the right than the left arm, because mental practice is based on feedforward process, since there is no movement feedback. Laterality effects in motor learning by mental practice may also emerge because the predictions of the nondominant arm control system would be relatively crude due to a lack of experience or use compared with the right arm (i.e., experience-dependent arm-dominance training). The previous theoretical considerations would predict that state estimation during mental actions should be more accurate and precise (i.e., more tightly related to actual state estimation) for the dominant arm than for the

nondominant arm, leading thus to better and faster motor learning for the former. In the current study, we aimed to investigate laterality effects in motor learning by motor-imagery practice. We asked four groups of right-hander adults to mentally and physically perform as fast and accurate as possible (speed/accuracy trade-off paradigm) successive reaching movements with their dominant or nondominant arm toward multiple targets following a pre-determined path. Based on the previously mentioned lateralization motor processes, we predicted that dominant arm mental training, compared to nondominant arm mental training, should result in higher enhancement of motor performance. As such, this study contributes to expand our knowledge regarding the underlying learning processes of the dominant and nondominant arms during mental practice.

EXPERIMENTAL PROCEDURES Participants Sixty healthy young adults participated in this experiment. All were right-handers (average score: 0.85 ± 0.04), as determined by the Edinburgh Handedness inventory (Oldfield, 1971), and good imagers (average score: 46 ± 3, maximum score 57) as determined by the French version of the Movement Imagery Questionnaire (Hall and Martin, 1997). Participants were randomly assigned into five groups (see Fig. 1): (i) the physicaltraining dominant arm group (Pd group, mean age 23.2 ± 2.0 yrs, five males and seven females), (ii) the mental training dominant arm group (Md group, mean age 23.3 ± 2.1 yrs, seven males and five females), (iii) the physical training nondominant arm group (Pnd group, mean age 24.1 ± 1.6 yrs, eight males and four females), (iv) the mental training nondominant arm group (Mnd group, mean age 22.8 ± 2.5 yrs, eight males and four females), and (v) the control-group (C group, mean age 22.1 ± 1.9 yrs, six males and six females). Handedness and imagery scores did not differ between groups (respectively, F(4,55) = 0.13, p = 0.97 and F(4,55) = 0.3, p = 0.99; between-subject one-way analysis of variance (ANOVA)). All the participants gave their informed consent. The experimental protocol was approved by the Ethics committee of the Universite´ de Bourgogne and carried out in agreement with legal requirements and international norms (Declaration of Helsinki, 1964). Experimental device and motor task The experiment device was similar to that used in the study of Gentili et al. (2010). Two aluminum dowels (length: 75 cm, diameter: 1 cm) were fixed on a vertical bar (height: 86 cm, width: 10 cm) 44 cm one above the other. On each dowel, we symmetrically placed four targets, two on the left and two on the right side of the vertical bar (Fig. 2). The horizontal distance that separated the near (3, 4, 5, 6) and farther (1, 2, 7, 8) targets from the vertical bar was 10 cm and 35 cm, respectively. The eight targets were switches (diameter of 5 mm) and were all linked to an electronic stopwatch. Another target (target

R. J. Gentili, C. Papaxanthis / Neuroscience 297 (2015) 231–242

233

Fig. 1. Design of experiment. The three sessions (pretest, training, posttest) for which the temporal performance was recorded and analyzed and the five training groups. Pd: physical dominant arm; Md: mental dominant arm; Pnd: physical nondominant arm; Mnd: mental nondominant arm; C: control.

Fig. 2. Experimental setup. The apparatus was composed of a vertical bar that supported two parallel dowels on which eight switches (T1. . .T8) were fixed. An additional switch used to mark the start and the end of the entire movement was placed in front of the vertical bar (T0). All the switches were connected to a stopwatch to record the movement time. A label was placed next to each switch in order to define the left (left panel) and right (right panel) paths that were mirrored. Four groups of right-handed participants physically and mentally performed reaching movement toward the targets of the right and left path. For conventions see Fig. 1.

0) was also linked to the same stopwatch and was placed on the table 20 cm from the base of the vertical bar. The initial pressure on the switch T0 triggered the stopwatch, while the last pressure stopped it, recording so the total movement time. Pressures on the other switches (T1. . .T8) allowed to record the time elapsed between two successive pressures (i.e., intermediate movement times). Participants were seated in front of the device at a distance corresponding to 70% of their arm’s length and had to follow the path conforming to their training arm: we called ‘right-path’ the targets followed by the Pd and Md groups and ‘left-path’ the targets followed by the Pnd and Mnd groups. In each path, a label was placed near each switch to indicate the order of the switches to be pointed (see Fig. 2). Note that paths were mirror images and therefore imposed identical dynamical constraints during the motion of the right or the left arm. The movement sequence started from- and ended- at the target T0. All participants had to press successively the following switches: 0, 1, 2, 3, 4, 5, 6, 7, 8, 1, 7, 0 (a sequence of 11 arm movements, i.e., 12 pressures). This movement sequence constituted a trial. Note also that participants had previously memorized the order of the targets composing the paths (see ‘familiarization

session’ below) and that all targets were visible during the whole trial. Therefore, our task emphasized improvement in movement speed from target-to-target and it was not a serial reaction time task requiring learning of sequence order. Experimental procedures and data acquisition The protocol comprised a familiarization session, a pretest session (three trials), a training session (60 trials), and a posttest session (three trials). Familiarization session. Before the experiment, all participants familiarized themselves with the set-up by pressing the switches with their dominant (Pd and Md groups) or nondominant (Pnd and Mnd groups) index fingertips in a random order (twice per switch). To rule out any potential training effects that could improve the speed of arm movements, we asked participants to move their right or left arm along the path very slowly. Subsequently, we asked them to memorize the spatial disposition of the switches composing the pointing path. After eight to ten practice runs, all participants had memorized the specific path (i.e., they were able to

234

R. J. Gentili, C. Papaxanthis / Neuroscience 297 (2015) 231–242

mentally represent the pointing paths and targets’ position with their eyes closed). Pretest session. The pretest was identical for all the participants. Precisely, they were requested to physically point as accurately and as fast as possible toward the 12 switches composing the pointing paths. The Pd and Md groups pointed with their right arm, the Pnd and Mnd groups pointed with their left arm, while for the C group half of the participants pointed with their right arm and half with their left arm. Participants were informed that all targets had to be reached in the specified order. As a consequence, if they missed a target they had to point again this target and then to continue with the other targets. The three pretest trials were separated by a 10-s inter-trial interval and no information concerning motor performance (i.e., movement time or velocity) was provided to participants during or after the pretest. The total time, the intermediate times (i.e., the time of each arm movement from target-to-target), and the number of the missed targets were measured for each participant and used as a baseline motor performance. Training session. During the training session, participants in the Pd and Pnd groups physically followed the pointing paths with their right and left arms, respectively. They were encouraged to move as accurately and as fast as possible at each trial. Individual motor performance (i.e., total movement time) was recorded for each trial. Participants in the Md and Mnd groups mentally simulated (kinaesthetic or first person imagery, as in Gentili et al., 2010) the pointing paths with their right and left arms, respectively. Imagining a movement at the first person is a necessary condition to engage the motor system (see Stinear et al., 2006b). They were encouraged at each trial to mentally move their arm along the pointing paths as accurately and as fast as possible, as they would do if they had to physically perform the task. The total movement time during mental training was also recorded for each trial by asking the participants to press the T0 target-switch with the right or left index just before starting the mental movements and to re-press it immediately after they had mentally accomplished the pointing path. By means of this method, we quantified the trial-by-trial temporal improvement during mental training and compared it with the temporal improvement during physical training. Participants in the C group were instructed to make accurate and quick gaze shifts to the switches composing the pointing path without moving their arms. We included this control group since we observed that the eyes of the participants of the other groups shifted from one switch to the other when they physically or mentally carried out the motor task. As the task required eye–hand coordination, improvement in arm motor performance (i.e., faster pointing movements) after mental training could be attributed to eye movement training, to an attention effect, or to an improvement in memorization of target locations, rather than to enhancement in motor performance per se. Eye movements are involved in mental actions (Gueugneau

et al., 2008; Heremans et al., 2008) and eye position signals are known to influence the cortical reach related network (Andersen and Buneo, 2002) and therefore to contribute to the neural commands to the limb. With this control task, we hoped to isolate a direct positive effect of mental training on motor performance from indirect effects. Posttest session. The posttest was identical to the pretest and given 2 min after the training session. The total time, the temporal gain, the mental/physical gain ratio for each hand modality, the intermediate times (i.e., the time of each arm movement from targetto-target), the intermediate mental/physical gain ratio (i.e., the mental/physical gain ratio of each arm movement from target-to-target for each hand modality) and the number of the missed targets, were measured for each participant and used as indexes of motor performance improvement. Statistical analysis Comparisons between pretest and posttest. All variables showed normal distribution (Shapiro–Wilk tests, p > 0.05) and equivalent variance (Hartley, Cochran and Bartlett tests, p > 0.05). Statistical effects for the total movement time were quantified by a 2 Modality  2 Hand  2 Test ANOVA with Modality (physical, mental) and Hand (dominant, nondominant) as between-subject factors and Test (Pretest, Posttest) as within-subject factor. Also, a two-tailed t-test (paired samples) was employed to assess differences between the pretest and posttest for the control group. The temporal gain (also called speed gain) between pretest and posttest was calculated as follows: gain ¼

pretest  postest  100 pretest

As for the movement times in pretest and posttest, the normality of the speed gain was assessed and the results revealed that these data were normally distributed (Shapiro–Wilk tests, p > 0.05). Statistical effects for the speed gain were accessed by means of a 2 Modality  2 Hand ANOVA with Modality (physical, mental) and Hand (dominant, nondominant) as between-subject factor.  When necessary post hoc comparisons were carried out by means of Tukey’s tests. Statistical differences in intermediate movement times between pretest and posttest were quantified by two-tailed t-tests (paired samples) for each group separately. To assess the differences between the dominant and the non-dominant arms on motor learning by mental practice, we calculated the ratio (Mental speed gain/Physical speed gain) of the speed gain (see above) between mental practice and   It must be noted that the speed gain (ratio of two normal variables) corresponds to the heavy-tailed, symmetric and bell-shaped Cauchy distribution. Thus, to further ensure the normality of these data, the speed gain was also assessed using a general linear model and the normality of the residual was analyzed. Consistent with the results from the Shapiro–Wilk test, the findings revealed that the residuals were normally distributed.

R. J. Gentili, C. Papaxanthis / Neuroscience 297 (2015) 231–242

physical practice for each arm. This ratio normalizes the speed gain of mental training with respect to the speed gain of physical practice and thus allows a direct comparison of motor improvement by means of mental practice between the dominant and non-dominant arms. Two-tailed t-tests (independent samples) were employed to assess any statistical differences between those two ratios. The number of missed targets in pretest and posttest did not show normal distribution (Shapiro–Wilk tests, p < 0.05). Therefore, statistical effects were quantified by means of Kruskal–Wallis ANOVA for between-group comparisons (Pd, Md, Pnd, Mnd, and C) and by means of Wilcoxon tests for within-group comparisons (Pretest versus Posttest). For all statistical analyses, significance was accepted at p < 0.05. Trial-by-trial analysis in the posttest. To investigate any trial-by-trial improvement in the posttest performance, finding which would indicate that the training session was incomplete, we performed a oneway ANOVA analysis (three posttest trials) for each group separately (Pd, Md, Pnd and Mnd). Post hoc comparisons were carried out by means of Tukey’s tests (level of significance, p < 0.05). Data fitting analysis of temporal performance in the training session. We quantified the trial-by-trial improvement in movement time during the training session with simple learning models. Specifically, we tested three possible models, in which the total movement time monotonically decreases with the number of trials in a negatively accelerated fashion: Hyperbolic : TimeðTrÞ ¼ a=ð1 þ b TrÞ þ c;

ð1Þ

Exponential : TimeðTrÞ ¼ a expðb TrÞ þ c;

ð2Þ

and Power : TimeðTrÞ ¼ a Trb þ c;

ð3Þ

where Time is the total movement time, a is the total amount of learning as the number of trials Tr tends to infinity, b is the learning rate, c is the asymptotic performance as the number of trials Tr tends to infinity. We found that the exponential model fitted the data well in all conditions and for all the participants (see Results section). In addition, we calculated the within-subject trial-by-trial variability by computing the root mean square of the error (rMSE) of the residuals between the exponential model and the data for each participant. The parameters a, b, c and rMSE showed normal distribution (Shapiro–Wilk tests, p > 0.05) and equivalent variance (Hartley, Cochran and Bartlett tests, p > 0.05). Statistical effects were accessed by means of a one-way ANOVA with group (Pd, Md, Pnd, Mnd, and C) as between-subject factor (level of significance, p < 0.05). Post hoc comparisons were carried out by means of Tukey’s tests. EMG recording and analysis during mental training Our main aim was to control that arm muscles were quiescent during mental training (Md and Mnd groups).

235

EMG signals were recorded from the Anterior Deltoid (AD), Medialis Deltoid (MD), Posterior Deltoid (PD), Pectoral Major Superior (PMS), Triceps Brachii (TB), and Biceps Brachii (BB) for both arms. EMG signals from the same muscles were also recorded at rest (five trials, 10 s per trial) after asking participants of the Md and Mnd groups to totally relax their muscles. Two silver-chloride surface electrodes of 10-mm diameter each were placed on the muscle belly (the skin has been previously shaved and cleaned) with an interelectrode distance (center to center) of 2 cm. The reference electrode was positioned on the lower side of the right wrist. EMG signals were recorded at a frequency of 1000 Hz, band-pass filtered (10–600 Hz) and stored for off-line analysis using BIOPACÒ software acquisition. EMG patterns were quantified by computing the Root Mean Square (RMS) during mental training (60 trials) and rest (average value of five trials) using the following formula: sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Z MD 1 RMS ¼ ðEMGÞ2 dt; MD 0

where MD is the movement duration. For each muscle, a one-way ANOVA was applied (one rest and 60 trials) to statistically access (level of significance, p < 0.05) any difference in muscle activation between the rest condition and the mental trials.

RESULTS Spatial accuracy Equivalent spatial precision between groups and tests was a preliminary condition before comparing the improvement in movement time after physical and mental training. Considering that participants of each group pointed 396 times toward the targets (three trials  11 movements  12 participants) in the pretest and 396 times in the posttest sessions, a small number of targets was missed ( 0.60). Improvement of movement time between pretest and posttest Fig. 3A depicts average values of the total movement time in pretest and posttest. An ANOVA revealed a significant interaction effect between Modality and Test (F(1,44) = 43.817, p < 0.0001) and between Hand and Test (F(1,44) = 16.267, p < 0.0003). Post hoc analyses revealed that, the total movement time obtained during the pretest was equivalent between the physical and mental practice (p > 0.98) as well as between the

236

R. J. Gentili, C. Papaxanthis / Neuroscience 297 (2015) 231–242

Fig. 3. Average movement duration, speed gain and mental/physical gain ratios. (A) Average movement duration (+standard deviation, S.D.) in the pretest (black bar) and posttest (white bar) sessions for the Pd, Md, Pnd and Mnd groups. Significance differences are illustrated by forks and stars. (B) Average speed gain (±S.D.) between the pretest and the posttest sessions for all groups. (C) Average speed gain ratios (±S.D.) between the mental and physical practice groups for the dominant and nondominant hands. For both the speed gain and speed gain ratio, significance differences are illustrated with forks and stars. ⁄⁄⁄p < 0.0001; ⁄⁄p < 0.001; ⁄p < 0.05. For conventions see Fig. 1.

dominant and nondominant hands (p > 0.05). In addition, the same post hoc analyses revealed that, compared to the pretest, the total movement time during the posttest was further reduced for the physical compared to the mental (p < 0.0002) practice as well as for the dominant compared to the nondominant (p < 0.0005) hands. Also, the total movement time obtained during the pretest and posttest was similar for the control group (p > 0.05). An ANOVA conducted for the speed gain revealed a main effect of Modality (F(1,44) = 55.992, p < 0.0001) and a main effect of Hand (F(1,44) = 30.444, p < 0.0001).à Post hoc analyses revealed that the gain was significantly higher for the actual compared to the mental practice (p < 0.0002) and for the dominant compared to the nondominant hand (p < 0.0002) (Fig. 3B). Importantly, the analysis of the ratio of speed gains (Fig. 3C) revealed better performance for the dominant arm compared to the nondominant arm (t(22) = 2.56; p = 0.02). Speed improvement between pretest and posttest was due to a consistent decrease in the time of all arm movements (i.e., intermediate times) composing the pointing sequence (compare black with gray curves in the Fig. 4A–D). T-tests comparisons between pretest and posttest revealed that all intermediate movement times significantly decreased in all groups (for all comparisons, t > 2.6, p < 0.02), except those between the targets 4–5 (t = 1.78, p = 0.10), the targets 5–6 (t = 1.95, p = 0.08), and the targets 1–7 in the Mnd group (t = 1.85, p = 0.09). It is also of interest that the pattern of temporal performance (i.e., the target-bytarget variations in movement durations) in the pretest was qualitatively similar to those in the posttest (compare black with gray curves). The coefficient correlations between the pretest and the posttest were 0.89, 0.94, 0.90 and 0.91 (in all cases, t > 5,

à It must be noted that, consistent with the ANOVA analysis, results from the general linear model applied to the speed gain also revealed main significant effects for both Hand and Modality (p < 0.001).

Fig. 4. Intermediate value of movement duration and mental/physical gain ratios. (A–D) Average values (+S.D.) of intermediate movement duration (duration elapsed between two successive targets) in the pretest (black line) and posttest (gray line) sessions for all groups. (E) Average values (+S.D.) of speed gain ratio between the mental and physical practice group for the dominant and nondominant hands. For conventions see Fig. 1.

p < 0.001) for the Pd, Md, Pnd and Mnd groups, respectively. Interestingly, the speed gain ratio between the dominant and nondominant hands, previously observed for the entire path (see Fig. 4E), was significantly larger for the dominant compared to the nondominant arm for

237

R. J. Gentili, C. Papaxanthis / Neuroscience 297 (2015) 231–242

arm movements between the targets 4 and 5 (t = 3.23, p = 0.004), targets 5 and 6 (t = 2.51, p = 0.02), targets 7 and 8 (t = 2.67, p = 0.01), as well as between targets 7 and 1 (t = 2.19, p = 0.04), (see Fig. 4E). Improvement of movement time during the training session Overall, the trial-by-trial analysis during the training session revealed that total movement time progressively decreased for all groups (see Fig. 5). We found that the exponential model fitted the data well in all conditions (on average 8% better than the two other models) and for all the participants (see, average R2 in the Table 1). Since for the four groups the training sessions generated similar learning curves, it was possible to directly compare the parameters of the exponential model used to fit the data. Table 1 shows average values for the parameters of the exponential model. ANOVA revealed a significant effect of group for all the parameters. For the parameter a (F(3,44) = 9.41, p < 0.001), which captures the amount of learning, post hoc comparisons showed significant differences

between the Mnd group and the three other groups (for all cases, p < 0.01). The other comparisons between groups did not reach the level of significance (in all cases, p > 0.2). For the learning rate b (F(3,44) = 5.41, p = 0.003), significance was found between the group Pd and the groups Md (p = 0.004) and Mnd (p = 0.009). The other comparisons between groups did not show any differences (in all cases, p > 0.20). Lastly, for the parameter c (F(3,44) = 20.13, p < 0.001), which captures asymptotic performance (i.e., the predicted value of performance after a large number of trials), differences were found between the Pd group and the Pnd (p = 0.003) and Mnd (p < 0.001) groups, between the groups Md and Mnd (p < 0.001), and between the groups Pnd and Mnd (p = 0.003). The analysis of variability (see rMSE in Table 1) showed also a significant effect of group (F(3,44) = 5.29, p = 0.003). Post hoc comparisons showed significant differences between the Mnd group and the three other groups (in all cases, p < 0.02). The other comparisons did not reach the level of significance (in all cases, p > 0.90). Note that for the mental trials of the Md and Mnd groups, we verified with surface EMG analysis that arm

Fig. 5. Average (+S.E.) learning curves during physical and motor-imagery practice with the dominant and the nondominant arms over the 60 practiced trials. For conventions see Fig. 1.

Table 1. Average parameters of the fit to model the learning curves for the four training groups. R2, is the coefficient of determination; a, is the total amount of learning; b, is the learning rate; c, is the asymptotic performance as the number of trials Tr tends to infinity; rMSE, is the within-subject trialby-trial variability by computing the root mean square of the error of the residuals between the exponential model and the data for each participant. For conventions see Fig. 1 Groups 2

R a b c rMSE

Pd

Md

Pnd

Mnd

0.82 ± 0.02 2.69 ± 0.13 0.20 ± 0.04 6.19 ± 0.08 3.11 ± 0.35

0.82 ± 0.03 2.33 ± 0.15 0.08 ± 0.01 6.65 ± 0.16 3.63 ± 0.44

0.78 ± 0.03 2.34 ± 0.13 0.13 ± 0.02 7.15 ± 0.11 3.65 ± 0.72

0.62 ± 0.06 1.70 ± 0.10 0.09 ± 0.01 8.09 ± 0.27 6.72 ± 1.02

238

R. J. Gentili, C. Papaxanthis / Neuroscience 297 (2015) 231–242

muscles remained silent. Precisely, RMS values during mental training in both groups were low (from 6 lV to 11 lV) and not statistically different (for each muscle separately; p > 0.50) to those recorded in the baseline condition (from 4 lV to 9 lV). Temporal performance during the posttests After physical training (see Pd and Pnd in Fig. 6), the temporal performance in posttest trials remained almost stable; the duration of the three posttest trials were not statistically different (F(2,22) = 1.63, p = 0.21 for Pd; F(2,22) = 2.10, p = 0.14 for Pnd). After mental training, however, a significant improvement in temporal performance was detected. For the Md group (see Md in Fig. 6) a marginal decrease in movement time was found between the 1st and 3rd posttest trials (F(2,22) = 4.27, p = 0.026). For the Mnd group, the decrease in movement time was highly significant (F(2,22) = 21.38, p < 0.0001) and between all the posttest trials (for all comparisons, p < 0.02).

DISCUSSION We investigated the laterality effects in motor learning resulting from physical and motor-imagery practice in right-handers by employing an ecologically valid 3D reaching task toward multiple targets (speed/accuracy trade-off paradigm). Qualitatively, we found that both types of training had a positive impact on motor performance of the dominant arm and the nondominant arm through similar learning process (i.e., similar exponential learning curves). Quantitatively, physical

practice was superior to motor-imagery practice for both arms. More importantly, motor-imagery practice with the dominant arm resulted in larger and more robust motor improvements than motor-imagery practice with the nondominant arm. Overall our findings suggest a superiority of the dominant arm in motor learning by employing mental practice.

Physical practice with the dominant and the nondominant arm Although physical training enhanced motor performance in both groups, motor improvement was larger for the dominant arm (Pd) compared to the nondominant arm (Pnd), as revealed by a higher reduction in movement time and differences in asymptotic performance during training (parameter c). A possible explanation would be that these laterality effects result from the relative complementary functional specialization of the cerebral hemispheres in motor control and learning. The left hemisphere would be predominantly involved in movement planning, prediction, and learning (e.g., Wolpert et al., 1998; Kuhtz-Buschbeck et al., 2003; Agnew et al., 2004), while primarily acting as a controller that predicts and optimizes arm/task dynamics (dynamic dominance-hypothesis theory; Sainburg, 2002; Mutha et al., 2012). Conversely, the right hemisphere, with more limited planning capabilities, would mainly encode general motion features (e.g., Ivry and Robertson, 1998; Rushworth et al., 2001; Grafton et al., 2002) further relying thus on peripheral feedback-mediated impedance control mechanisms to reach stable positions (Sainburg, 2002; Mutha et al., 2012). In our task, although final position was important, performance improvements mainly depended on increasing arm speed in order to reach as fast as possible each target composing the path. Thus, a trial-by-trial speed increase required the precise control of arm dynamics (i.e., gravity, inertial and interactions torques), which substantially varied during arm reaching motion. Thus, as the dominant arm controller has expertise in controlling arm dynamics, the Pd group further benefitted from physical practice than the Pnd group.

Mental practice with the dominant and the nondominant arm

Fig. 6. Average values (+S.D.) of the three trials (T1, T2 and T3) in the posttest session for all the groups. Significance differences are illustrated with forks and stars. ⁄⁄⁄p < 0.0001; ⁄⁄p < 0.001; ⁄ p < 0.05. For conventions see Fig. 1.

Reduction of movement time in the posttest session for almost all the targets composing the pointing path suggests a positive impact of motor-imagery training in the motor performance of both Md and Mnd groups. This is consistent with results from previous studies, which have shown beneficial effects of motor-imagery practice on motor learning and performance (Yue and Cole, 1992; Ya´gu¨ez et al., 1998; Ranganathan et al., 2004; Gentili et al., 2006, 2010; Allami et al., 2008; Debarnot et al., 2009, 2011). Overall, our findings confirm and extend those from previous studies (e.g., Yue and Cole, 1992; Page et al., 2007; Gandrey et al., 2013) suggesting that the nondominant arm can benefit from motor-imagery training during reaching movements involving a speed-accuracy trade-off reinforcing thus the

R. J. Gentili, C. Papaxanthis / Neuroscience 297 (2015) 231–242

idea that mental practice can be generalized successfully to different body segments and task constraints. Before going ahead, alternative interpretations must be eliminated. First, motor improvement can be reasonably attributed to mental practice itself since no muscle activity was detected during the entire training. Second, due to similar movement precision in pretest and posttest, performance cannot be ascribed to a strategy rather based on speed than accuracy. Third, since the performance of the control group was unchanged, motor improvement in Md and Mnd groups cannot be explained in terms of non-motor phenomena (e.g., attention, eye movement training). Finally, motor improvement could not be due to learning the target order since (i) the targets were fixed and clearly identified (no reaction time requirement) and (ii) the control group that could have benefited of such learning (eye movement training) kept its performance unchanged. Comparisons between physical and mental practice By analyzing trial-by-trial the mental and actual movement time during the training sessions, we found that the learning curves during motor-imagery (Md and Mnd group) and physical training (Pd and Pnd groups) follow similar exponential shapes (see also Gentili et al., 2010). This finding may suggest that physical and motor-imagery trainings share a common plastic learning process. A possible underlying mechanism for these qualitative similarities in motor learning can be provided by the internal model theory (Wolpert and Miall, 1996; Wolpert and Ghahramani, 2000; Wolpert and Flanagan, 2001). Briefly, during physical practice the internal forward model receives as input a copy of the ongoing motor commands and the initial state of the arm in order to predict the future states of the arm (e.g., position, time, velocity). During mental practice motor commands are prepared without reaching the muscular level. Therefore, a copy of the motor commands and the initial state of the arm are still available to the forward model, which generate predictions for the future states of the arm. We propose that state simulation, based on forward internal model output, may be the common process during physical and mental practice, which guides motor performance improvement. Although we found qualitative similarities between the physical training and the motor-imagery training learning curves, there were notable quantitative differences. For instance, the learning rate (parameter b) in motorimagery training was smaller than the learning rate obtained in physical training. As a result, the total amount learned at the end of training (parameter a) was less in motor-imagery training than in physical training. Additionally, there were important differences in the asymptotic performance (parameter c) between physical and motor-imagery training. These results probably suggest that state estimation during motor-imagery training is more variable than during physical training. Because of the trial-by-trial variability in the state estimation, the internal training signal used for learning is itself variable, presumably resulting in a slower update of the motor controller during motor-imagery training compared to physical training where the peripheral

239

feedback is available. This idea is further corroborated by the finding that movement speed still improved in the posttest session for both the Md and the Mnd groups, but not for the Pd and Pnd groups. This may indicate that mental training sessions were incomplete and that actual movement production following motor-imagery practice could further improve motor performance due to the update of state estimation by sensory information provided during movement execution. Dominant arm versus nondominant arm differences during and after mental practice We found substantial differences in motor learning and performance by motor-imagery practice between the dominant (Md group) and the nondominant (Mnd group) arm. First, movement time reduction after the mental training session was about 1.5 times higher for the dominant than the nondominant arm. In addition, the ratio of speed gain suggests that the improvement in performance after a mental training conducted with the dominant arm was higher compared to the nondominant arm (see Figs. 3C, 4E). Importantly, this laterality effect was not driven by marginal gains, but was consistent in all arm movements composing the pointing paths. Second, the learning process, although qualitatively similar between the two arms (exponentials learning curves), showed an advantage for the dominant arm: the amount of learning (parameter a) and the asymptotic performance (parameter c) were greater for the Md than the Mnd group. Furthermore, variability was larger in the Mnd compared to the Md group. Third, motor performance in the posttest session was more stable in the Md than the Mnd group. Indeed, there was an important trial-by-trial reduction in movement time for the Mnd group during the three posttest trials, while a marginal reduction was found in movement time between the first and third posttest trials in the Md group. Overall, motor-imagery practice with the dominant arm could be considered as more efficient than motor-imagery practice with the nondominant arm. Note that as movement paths for both arms were mirror images, precision and dynamic constraints were theoretically identical, and therefore laterality effects cannot be attributed to different task constraints. The computational framework of internal forward model combined with the dynamic-dominance hypothesis for laterality could provide a possible explanation to the observed laterality effects. This latter proposes a hybrid control scheme in which both arms could beneficiate from predictive and feedback control mechanisms at different degrees (Yadav and Sainburg, 2011): the dominant/left hemisphere system would rely more on predictive control mechanisms, while the nondominant/right hemisphere system would rely more on feedback-mediated impedance control mechanisms (e.g., Sainburg, 2002; Mutha et al., 2013). As previously mentioned, performance improvements in our 3D reaching task predominantly required the control of complex arm dynamics. As such the dominant arm, which would be guided by efficient predictive mechanisms (e.g., Wolpert et al., 1998; Kuhtz-Buschbeck et al., 2003;

240

R. J. Gentili, C. Papaxanthis / Neuroscience 297 (2015) 231–242

Agnew et al., 2004; Mutha et al., 2011), seems to be better suited to take advantage from a motor-imagery training session. Specifically, accurate and precise state estimation during motor-imagery training with the dominant arm could be used to train the controller (by computing an internal error signal) and enhance the neural commands to improve motor performance (Jordan and Rumelhart, 1992). Conversely, the nondominant arm controller, which would further rely on feedback control, would be less efficient to enhance the performance since no peripheral feedback is available during mental practice. For that reason, in our study the performance enhancement in the Mnd group was inferior compared to all other groups. However, if the nondominant arm controller was only involved in feedback control no enhancement should have been observed at all. Likely, the nondominant arm controller has access to predictive mechanisms. Though, state estimation should be less accurate and noisy, resulting thus to modest and less steady performance improvement. It must be noted that such hemisphere specialization may be attenuated or even reverse based on the nature of the task demand itself. Namely, considering a task that would emphasize of final accuracy and less the integration of the mechanical constraints related to arm dynamics, it could be expected that the nondominant arm controller would be equally or better suited to perform. Future experiments that will manipulate the nature of reaching tasks performed during mental practice may help to further answer this question. However, other potential not exclusive neural mechanisms could also account for the laterality effects observed during mental execution. Specifically, both arm controllers could include and utilize a forward model during mental practice. However, the forward model of the nondominant arm controller would be relatively coarse possibly due to a lack of previous motor experience with the nondominant arm. Thus, such a crude forward model would generate noisy predictive signals which in turn would be less effective to guide motor learning during mental practice. As a result, in order to compensate such uncertainty of the predictions the nondominant arm controller would need to further rely on peripheral feedback to allow its forward model to refine its predictions as well as to guide arm movements. This would result in a weaker overlap of the neural control strategies between mental and actual practice executed with the nondominant arm increasing thus performance discrepancy between these two execution modalities. The observed laterality effects may also be a combination of these different neural mechanisms (i.e., arm-dynamic hypothesis and experience-dependent arm-dominance training). The specifications of these neural mechanisms and their relationships are somewhat speculative at this point and thus further investigation is needed. Note that the superiority of the dominant arm for motor learning by motor-imagery practice was not the only possible outcome in our study. Instead, one could expect equal motor improvement for both arms after motor-imagery training as several neuroimaging studies have shown bilateral activation of primary motor cortex

during imagined actions (Lotze et al., 1999; Lacourse et al., 2005; Creem-Regehr et al., 2007; Pelgrims et al., 2011). These findings would support the premise that similar motor plans and predictions are shared by the two sides of the body. Notably, it has been recently reported that the effects of virtual lesions of left and right primary motor cortices in individuals performing a mental rotation task were the same irrespective of the laterality (left/right) of hand drawings (Pelgrims et al., 2011). Our findings suggest that beyond the activation of similar neural networks, motor-imagery practice with the dominant arm, compared to the nondominant arm, may engage more accurate predictive mechanisms that better promote motor learning. Conceptually, the idea of an amodal representation of motor states, namely the premise that mental imagery process may involve the same form of representation as that of reasoning in general, with the exception that the content of thoughts experienced as images includes information about how things would look (Pylyshyn, 2002), could also challenge any laterality effect of motor learning by mental practice. Indeed, if during motor-imagery training an explicit knowledge is used to simulate what would happen if a movement was actually executed, without representing the kinematics and the dynamics of a particular movement and estimating the sensory consequences of the motor commands, there is no obvious reason to expect any laterality effects, and even any motor learning, after mental practice.

CONCLUSION This study reinforces the idea that mental and physical practice share common plastic neural mechanisms underlying motor learning and performance with the dominant and nondominant upper limb. Compared to the nondominant arm, the mental practice with the dominant arm results in higher performance improvements. Such differences may be based on a relative specialization (dynamic-dominance hypothesis) and/or motor experience (experience-dependent arm-dominance training) of the cerebral hemispheres. These mechanisms remain somewhat speculative and thus further examination is needed. Future work could examine if the neural processes discussed here apply to other types of motor tasks that are not sequential in nature and employ neuroimaging techniques to inform actual/mental performance to complement behavioral measurements. Overall, as both arms benefit from mental practice, this work can inform motor imagerybased practice for rehabilitation of the dominant/ nondominant upper extremity and thus expands the understanding of the hemisphere specificity during arm reaching movements from motor to mental states of action.

REFERENCES Agnew JA, Zeffiro TA, Eden GF (2004) Left hemisphere specialization for the control of voluntary movement rate. Neuroimage 22(1):289–303. Allami N, Paulignan Y, Brovelli A, Boussaoud D (2008) Visuo-motor learning with combination of different rates of motor imagery and physical practice. Exp Brain Res 184(1):105–113.

R. J. Gentili, C. Papaxanthis / Neuroscience 297 (2015) 231–242 Amunts K, Schlaug G, Schleicher A, Steinmetz H, Dabringhaus A, Roland PE, Zilles K (1996) Asymmetry in the human motor cortex and handedness. Neuroimage 4:216–222. Andersen RA, Buneo CA (2002) Intentional maps in posterior parietal cortex. Annu Rev Neurosci 25:189–220. Avanzino L, Bove M, Tacchino A, Trompetto C, Ogliastro C, Abbruzzese G (2009) Interaction between finger opposition movements and aftereffects of 1Hz-rTMS on ipsilateral motor cortex. J Neurophysiol 101(3):1690–1694. Bakker M, de Lange FP, Stevens JA, Toni I, Bloem BR (2007) Motor imagery of gait: a quantitative approach. Exp Brain Res 179(3):497–504. Collet C, Di Rienzo F, El Hoyek N, Guillot A (2013) Autonomic nervous system correlates in movement observation and motor imagery. Front Hum Neurosci 7:415. Creem-Regehr SH, Neil JA, Yeh HJ (2007) Neural correlates of two imagined egocentric transformations. Neuroimage 35(2):916–927. Debarnot U, Creveaux T, Collet C, Gemignani A, Massarelli R, Doyon J, Guillot A (2009) Sleep-related improvements in motor learning following mental practice. Brain Cogn 69(2):398–405. Debarnot U, Clerget E, Olivier E (2011) Role of the primary motor cortex in the early boost in performance following mental imagery training. PLoS One 6(10):e26717. Decety J, Jeannerod M (1995) Mentally simulated movements in virtual reality: does Fitts’s law hold in motor imagery? Behav Brain Res 72(1–2):127–134. Decety J, Jeannerod M, Durozard D, Baverel G (1993) Central activation of autonomic effectors during mental simulation of motor actions in man. J Physiol 461:549–563. Demougeot L, Normand H, Denise P, Papaxanthis C (2009) Discrete and effortful imagined movements do not specifically activate the autonomic nervous system. PLoS One 4(8):e6769. Desmurget M, Grafton S (2000) Forward modeling allows feedback control for fast reaching movements. Trends Cogn Sci 4(11):423–431. Fadiga L, Buccino G, Craighero Fogassi L, Gallese V, Pavesi G (1999) Corticospinal excitability is specifically modulated by motor imagery: a magnetic stimulation study. Neuropsychologia 37(2):147–158. Gandrey P, Paizis C, Karathanasis V, Gueugneau N, Papaxanthis C (2013) Dominant vs. nondominant arm advantage in mentally simulated actions in right handers. J Neurophysiol 110(12): 2887–2894. Gentili RJ, Cahouet V, Ballay Y, Papaxanthis C (2004) Inertial properties of the arm are accurately predicted during motor imagery. Behav Brain Res 155(2):231–239. Gentili RJ, Papaxanthis C, Pozzo T (2006) Improvement and generalization of arm motor performance through motor imagery practice. Neuroscience 137(3):761–772. Gentili RJ, Han CE, Schweighofer N, Papaxanthis C (2010) Motor learning without doing: trial-by-trial improvement in motor performance during mental training. J Neurophysiol 104(2): 774–783. Grafton ST, Hazeltine E, Ivry RB (2002) Motor sequence learning with the nondominant left hand. A PET functional imaging study. Exp Brain Res 146:369–378. Gueugneau N, Crognier L, Papaxanthis C (2008) The influence of eye movements on the temporal features of executed and imagined arm movements. Brain Res 1187:95–102. Guillot A, Collet C (2005) Contribution from neurophysiological and psychological methods to the study of motor imagery. Brain Res Brain Res Rev 50(2):387–397. Haaland KY, Harrington DL (1996) Hemispheric asymmetry of movement. Curr Opin Neurobiol 6:796–800. Hall CR, Martin KA (1997) Measuring movement imagery abilities: a revision of the Movement Imagery Questionnaire. J Mental Imagery 21(1–2):143–154. Heremans E, Helsen WF, Feys P (2008) The eyes as a mirror of our thoughts: quantification of motor imagery of goal-directed

241

movements through eye movement registration. Behav Brain Res 187(2):351–360. Ivry RB, Robertson LC (1998) The two sides of perception. Cambridge: MIT Press. Jackson PL, Lafleur MF, Malouin F, Richards CL, Doyon J (2003) Functional cerebral reorganization following motor sequence learning through mental practice with motor imagery. Neuroimage 20(2):1171–1180. Jeannerod M (2001) Neural simulation of action: a unifying mechanism for motor cognition. Neuroimage 14(1 Pt 2): S103–S109. Jordan MI, Rumelhart DE (1992) Forward models: supervised learning with a distal teacher. Cogn Sci 16(3):307–354. Kawato M (1999) Internal models for motor control and trajectory planning. Curr Opin Neurobiol 9(6):718–727. Kuhtz-Buschbeck JP, Mahnkopf C, Holzknecht C, Siebner H, Ulmer S, Jansen O (2003) Effector-independent representations of simple and complex imagined finger movements: a combined fMRI and TMS study. Eur J Neurosci 18(12):3375–3387. Lacourse MG, Orr ELR, Cramer SC, Cohen MJ (2005) Brain activation during execution and motor imagery of novel and skilled sequential hand movements. Neuroimage 27(3): 505–519. Lafleur MF, Jackson PL, Malouin F, Richards CL, Evans AC, Doyon J (2002) Motor learning produces parallel dynamic functional changes during the execution and imagination of sequential foot movements. Neuroimage 16(1):142–157. Lorey B, Bischoff M, Pilgramm S, Stark R, Munzert J, Zentgraf K (2009) The embodied nature of motor imagery: the influence of posture and perspective. Exp Brain Res 194(2):233–243. Lotze M, Montoya P, Erb M, Hu¨lsmann E, Flor H, Klose U, Birbaumer N, Grodd W (1999) Activation of cortical and cerebellar motor areas during executed and imagined hand movements: an fMRI study. J Cogn Neurosci 11(5):491–501. Maruff P, Wilson PH, De Fazio J, Cerritelli B, Hedt A, Currie J (1999) Asymmetries between dominant and non-dominant hands in real and imagined motor task performance. Neuropsychologia 37(3):379–384. Munzert J, Lorey B, Zentgraf K (2009) Cognitive motor processes: the role of motor imagery in the study of motor representations. Brain Res Rev 60(2):306–326. Mutha PK, Sainburg RL, Haaland KY (2011) Left parietal regions are critical for adaptive visuomotor control. J Neurosci 31(19):6972–6981. Mutha PK, Haaland KY, Sainburg RL (2012) The effects of brain lateralization on motor control and adaptation. J Motor Behav 44(6):455–469. Mutha PK, Haaland KY, Sainburg RL (2013) Rethinking motor lateralization: specialized but complementary mechanisms for motor control of each arm. PLoS One 8(3):e58582. Oldfield RC (1971) The assessment and analysis of handedness: the Edinburgh inventory. Neuropsychologia 9(1):97–113. Page SJ, Levine P, Leonard A (2007) Mental practice in chronic stroke: results of a randomized, placebo-controlled trial. Stroke 38(4):1293–1297. Page SJ, Szaflarski JP, Eliassen JC, Pan H, Cramer SC (2009) Cortical plasticity following motor skill learning during mental practice in stroke. Neurorehabil Neural Repair 23(4):382–388. Papaxanthis C, Schieppati M, Gentili R, Pozzo T (2002) Imagined and actual arm movements have similar durations when performed under different conditions of direction and mass. Exp Brain Res 143(4):447–452. Papaxanthis C, Paizis C, White O, Pozzo T, Stucchi N (2012) The relation between geometry and time in mental actions. PLoS One 7(11):e51191. Pelgrims B, Michaux N, Olivier E, Andres M (2011) Contribution of the primary motor cortex to motor imagery: a subthreshold TMS study. Hum Brain Mapp 32(9):1471–1482. Pylyshyn ZW (2002) Mental imagery: in search of a theory. Behav Brain Sci 25(2):157–182.

242

R. J. Gentili, C. Papaxanthis / Neuroscience 297 (2015) 231–242

Ranganathan VK, Siemionow V, Liu JZ, Sahgal V, Yue GH (2004) From mental power to muscle power – gaining strength by using the mind. Neuropsychologia 42(7):944–956. Rushworth MF, Krams M, Passingham RE (2001) The attentional role of the left parietal cortex: the distinct lateralization and localization of motor attention in the human brain. J Cogn Neurosci 13:698–710. Sabate´ M, Gonza´lez B, Rodrı´ guez M (2004) Brain lateralization of motor imagery: motor planning asymmetry as a cause of movement lateralization. Neuropsychologia 42(8):1041–1049. Sainburg RL (2002) Evidence for a dynamic-dominance hypothesis of handedness. Exp Brain Res 142(2):241–258. Schluter ND, Rushworth MF, Passingham RE, Mills KR (1998) Temporary interference in human lateral premotor cortex suggests dominance for the selection of movements. A study using transcranial magnetic stimulation. Brain 121:785–799. Serrien DJ, Ivry RB, Swinnen SP (2006) Dynamics of hemispheric specialization and integration in the context of motor control. Nat Rev Neurosci 7(2):160–166. Skoura X, Personnier P, Vinter A, Pozzo T, Papaxanthis C (2008) Decline in motor prediction in elderly subjects: right versus left arm differences in mentally simulated motor actions. Cortex 44(9):1271–1278. Skoura X, Vinter A, Papaxanthis C (2009) Mentally simulated motor actions in children. Dev Neuropsychol 34(3):356–367. Stinear CM, Fleming MK, Byblow WD (2006a) Lateralization of unimanual and bimanual motor imagery. Brain Res 1:139–147.

Stinear CM, Byblow WD, Steyvers M, Levin O, Swinnen SP (2006b) Kinesthetic, but not visual, motor imagery modulates corticomotor excitability. Exp Brain Res 168(1–2):157–164. Wolpert DM, Flanagan JR (2001) Motor prediction. Curr Biol 11(18):729–732. Wolpert DM, Ghahramani Z (2000) Computational principles of movement neuroscience. Nat Neurosci 3:1212–1217. Wolpert DM, Miall RC (1996) Forward models for physiological motor control. Neural Netw 9(8):1265–1279. Wolpert DM, Ghahramani Z, Jordan MI (1995) An internal model for sensorimotor integration. Science 269(5232):1880–1882. Wolpert DM, Goodbody SJ, Husain M (1998) Maintaining internal representations: the role of the human superior parietal lobe. Nat Neurosci 1(6):529–533. Yadav V, Sainburg RL (2011) Motor lateralization is characterized by a serial hybrid control scheme. Neuroscience 196:153–167. Ya´gu¨ez L, Nagel D, Hoffman H, Canavan AG, Wist E, Ho¨mberg V (1998) A mental route to motor learning: improving trajectorial kinematics through imagery training. Behav Brain Res 90(1):95–106. Yue G, Cole KJ (1992) Strength increases from the motor program: comparison of training with maximal voluntary and imagined muscle contractions. J Neurophysiol 67(5):1114–1123. Zijdewind I, Toering ST, Bessem B, Van Der Laan O, Diercks RL (2003) Effects of imagery motor training on torque production of ankle plantar flexor muscles. Muscle Nerve 28(2):168–173.

(Accepted 7 February 2015) (Available online 19 March 2015)