Short-Term Limb Immobilization Affects Motor Performance

Abstract

C. Ghez, J. Gordon, and M. F. Ghilardi (1995; J. Gordon, M. F. Ghilardi, & C. Ghez, 1995; R. L. Sainburg, M. F. Ghilardi, H. Poizner, & C. Ghez, 1995) have found that proprioceptive deafferentation impairs feedforward and feedback mechanisms that control reaching movements. In the present study, the authors found immobilization-induced changes in limb kinematics, including joint motion, in 32 healthy participants who performed out-and-back movements before and after 0, 6, or 12 hr of immobilization of the left arm. Control participants did not undergo the arm immobilization procedure. Immobilization for 12 hr, but not 6 hr, caused trajectories with increased hand-path areas and altered interjoint coordination. The abnormalities were smaller in amplitude but similar in quality to those reported in deafferented patients (R. L. Sainburg et al.). In addition, movement onset point significantly drifted after immobilization. Thus, short-term limb disuse can affect interjoint coordination by acting on feedforward mechanisms. These behavioral alterations are potentially related to cortical plastic changes.

Proprioceptive deafferentation in humans changes motor performance (Ghez, Gordon, & Ghilardi, 1995; Gordon, Ghilardi, & Ghez, 1995; Sainburg, Ghilardi, Poizner, & Ghez, 1995). In particular, deafferented patients display abnormal hand trajectories and altered interjoint coordination because loss of proprioception impairs both feedback and feedforward mechanisms that control movement planning and execution (Sainburg et al.). In fact, the proprioceptive input provides a feedback signal to produce online corrections during reaching movements (feedback mechanisms), but the nervous system also uses the input to form and update internal models or memories that it can use to program movements in advance (feedforward mechanisms). Internal models of limb dynamics seem to be particularly important for neural control of biomechanical interactions during reaching movements: Because of transmission delays, the use of online proprioceptive information would in fact produce significant trajectory deviations resulting from inadequate coordination of muscle actions with interaction torques (Sainburg et al.). In addition, it is very plausible that the nervous system uses proprioceptive, but not visual, information to specify the precise timing of muscle activity because visual input produces only transitory and incomplete compensation in instances of proprioceptive loss (Ghez et al.; Sainburg et al.).

Investigators have recently introduced limb disuse or immobilization (i.e., sensorimotor restriction that is functionally different from the complete deafferentation that follows nerve lesions or limb amputation) as an effective rehabilitation tool for a few neurological conditions, including rehabilitation after stroke (Liepert et al., 1998; Taub & Uswatt, 2006) and dystonia (Priori, Pesenti, Cappellari, Scarlato, & Barbieri, 2001). Therefore, it is surprising that kinematic effects related to limb disuse are not well studied. Upper limb immobilization induces synaptic remodeling in the corresponding sensory and motor cortical areas (Huber et al., 2006). Those changes, which result from fixed proprioceptive peripheral signals, may partially mimic an acute proprioceptive deafferentation. In turn, the changes could cause modification of sensorimotor memories and possibly affect the processing of proprioceptive signals.

Little is known about the length of the limb disuse period that is needed to induce changes in motor performance or neuroplastic changes in normal people. The first studies were conducted on participants who wore splints either at the upper limb or the lower limb for 4−6 weeks because of fractures (Kaneko, Murakami, Onari, Kurumadani, & Kawaguchi, 2003; Liepert, Tegenthoff, & Malin, 1995; Seki, Taniguchi, & Narusawa, 2001a, 2001b). Long-term joint immobilization is known to modify the proportion of fast and slow muscle fibers and thus the contractility of skeletal muscle (Bey et al., 2003; Seki et al., 2001a) and perhaps to change the intrinsic properties of motor neurons and the input to motor neurons from peripheral afferents (Seki et al., 2001b). Although the results of some electrophysiological studies have suggested that changes of neural excitability at both central and peripheral levels may occur only after weeks or even months of immobilization, immobilization indeed induces modification of the sensorimotor input and motor behavior as soon as the cast is applied. Thus, changes of neural excitability either at cortical or subcortical levels may occur much earlier than previously believed. More recently, investigators have shown that 4 days of motor restriction of two fingers induced a decrease of cortical excitability without affecting excitability of nerve, muscle, and spinal motor neurons (Facchini, Romani, Tinazzi, & Aglioti, 2002). Huber et al. (2006) found that cortical plastic changes may occur with a shorter period of immobilization: Upper limb immobilization for 12 hr induced significant changes in somatosensory-evoked potentials (SEPs), amplitude of motor-evoked potentials (MEPs), and slow wave activity during sleep over the sensorimotor areas corresponding to the immobilized limb. Most interesting, alterations of hand path formation that were reminiscent of those found in deafferented patients (Sainburg et al., 1995) accompanied those electrophysiological changes.

In the present study, we analyzed changes in the location of movement onset and alterations in hand and joint trajectories to further characterize immobilization-induced motor alterations. We also defined how long immobilization has to last to produce measurable kinematic changes by testing participants after 6 hr and 12 hr of immobilization. We hypothesized that during short-term immobilization, dynamic proprioceptive input would decrease in otherwise normal sensory pathways. Thus, when participants started moving again, their proprioceptive sensation would be reliable for feedback, but their internal models of limb representation would be changed, affecting feedforward regulation of movement. That change would result in movements with an abnormal hand path and decreased interjoint coordination. However, those abnormalities would be of lesser extent than those found in proprioceptively deafferented patients, whose feedback and feedforward mechanisms are both impaired.

Method

Participants

Forty-six right-handers (M age = 27 years, range = 23−39 years) participated in the study. They had normal neurological examination, normal vision, and no previous history of neurological or orthopedic problems. We obtained written informed consent from all participants. We conducted the experiments in accordance with the Declaration of Helsinki. All participants were naive to our purpose in the experiment. In Experiment 1A, we assigned 21 participants to one of three groups (n = 7 in each). We immobilized the left arm of one group of participants for 12 hr and the left arm of another group of participants for 6 hr. We did not immobilize the arm of participants in the third group (controls). In Experiment 1B, we immobilized the left arm of 18 participants for 12 hr. We did not immobilize the arm of 7 participants, who served as controls.

Procedure

Immobilization procedure

We used a plastic splint to immobilize the left upper limb. The shape of the splint enabled it to keep the elbow joint at 90° of rotation, preventing any movement of the elbow joint. We fixed the splint to the arm with Velcro straps, and we supported the immobilized left limb with a sling to minimize shoulder movements.

Experimental procedure

Participants sat in front of a computer screen, with their left forearm supported over a digitizing tablet by a Plexiglas sled and with their left hand and forearm in a light cast that prevented wrist movements. We attached a magnetic cursor to the front of the sled in correspondence to the fingertips. We adjusted the height of the seat so that the arm moved in a horizontal plane at shoulder level. At the starting point, elbow and shoulder joint angles were 90° and 45°, respectively (see Figure 1A). Belts mechanically restrained movements of the torso and the shoulder. Moreover, one of the experimenters monitored participants' shoulder movements during data collection. An opaque shield prevented participants from viewing their hand and the tablet during the experiments.

FIGURE 1.

(A) Shoulder (S) and elbow (E) angles at beginning of each block. When movement onset was located in the center of the tablet, the angles measured 45° and 90°, respectively. (B) Representative hand paths for one participant at baseline, with and without visual feedback, to targets at 45°, 90°, and 135°. FB = feedback.

Targets were displayed on a computer screen at 5.8 cm from a common starting point at the 45°, 90°, and 135° directions (see Figure 1B). We presented the three targets pseudorandomly at 1.8-s intervals in blocks of eight cycles of six targets (two repetitions for each target), for a total of 48 presentations. We instructed participants to make straight, uncorrected out-and-back movements, with a sharp and fast reversal in the target circle. After familiarization with the experimental setup, the cursor was visible during movements in the first baseline block (visual baseline). In the following block, after the initial cursor alignment in the central starting position, we withheld cursor feedback, and participants performed movements without visual feedback (baseline). Then, according to the experimental protocol in Experiment 1A, we immobilized 7 participants for 12 hr and 7 participants for 6 hr. During that time interval, we allowed participants to perform their daily activities. When the splint was removed, we tested participants with a block of movements without visual feedback of the cursor (test). We did not immobilize control participants, and we tested them 12 hr later with the same protocol (without visual feedback).

We conducted Experiment 1B to verify the effect of visual feedback on motor performance. To ensure proper realignment in this experiment, we tested 25 different participants. To ensure proper realignment of the cursor, we provided participants with visual feedback once every three movements, before movement onset. We tested 18 participants at baseline and after 12-hr immobilization. The arm of each of the remaining 7 participants was not immobilized; they served as controls.

Data collection and analysis

We sampled hand trajectories at 200 Hz and collected the trajectories with custom-designed software. We computed several parameters for each movement, including (a) movement extent (i.e., length of the vector from the start to the end point), (b) amplitude of peak velocity, (c) directional error (i.e., directional difference between the vector from the start to the target and the vector from the start to the point of peak velocity; Ghilardi et al., 2000; Ghilardi, Eidelberg, Silvestri, & Ghez, 2003; see Table 1), and (d) normalized hand-path area (an index of hand-path shape, measured as the area enclosed by the hand path divided by the squared movement length; Huber et al., 2006).

TABLE 1.

Mean (± Standard Error) Movement Extent, Peak Velocity, and Directional Error in Baseline and Test Conditions for Unrestrained Controls and 6-hr and 12-hr Immobilized (IMM) Participants

	Controls		6-hr IMM		12-hr IMM
Condition	M	SE	M	SE	M	SE
Movement extent (cm)
Baseline	11.0	1.0	10.3	1.2	9.4	1.7
Test	11.7	1.0	10.4	1.3	8.6	0.9
Peak velocity (cm/s)
Baseline	52.5	6.6	44.1	5.7	44.6	4.7
Test	51.3	5.4	42.8	4.5	42.7	6.1
Directional error (deg)
Baseline	2.7	1.4	2.0	2.0	2.4	1.1
Test	4.7	1.5	1.9	2.0	0.5	1.8

We estimated joint movements from hand trajectories by using measurements of upper limb segments of each participant and assuming the shoulder position to be fixed. We computed interjoint timing (IJT) from the derived joint trajectory as the time interval between elbow and shoulder reversals (Sainburg et al., 1995). We derived IJTs only for all groups of participants in Experiment 1A because we did not measure limb segments lengths of participants in Experiment 1B.

To characterize the location changes of the movement starting point during trial blocks, we calculated the average x, y coordinates of the onset points every six movements (i.e., one cycle); in other words, we calculated the average of every six movements (i.e., Movements 1−6, 7−12, and so on). We then derived the Euclidean distance of the average from the original onset point and the corresponding angle of elbow and shoulder joints. We then measured drifts as the difference between those new points and the initial ones (see Figure 1A).

Statistical analysis

We submitted most dependent measures reported here to a mixed-model analysis of variance (ANOVA; α = .05) with block as a within-factor and group as a between-factor. We analyzed significant main effects and interactions further by using Bonferroni-corrected post hoc tests. We analyzed the distribution of the onset points (x,y coordinates) by using multivariate ANOVA with block as a within-factor and group as a between-factor.

Results

At baseline, with visual feedback (visual baseline), all participants executed straight and accurate movements, with sharp reversals and overlapping out-and-back strokes and without a difference between targets (see Figure 1B). When participants performed the same reaching task without visual feedback (baseline), movements were less accurate and slightly hypermetric, with small directional errors (see Figure 1B), as indicated in previous reports (Ghilardi et al., 2000; Ghilardi et al., 2003; see Table 1). In addition, over successive movements the location of movement onset drifted away from the visible starting point (see Figure 2A). At the end of the block, the distance of movement onset from the visible starting point was approximately 5 cm and was similar in all groups (M ± SE of the last six movements for no-immobilization, 6-hr-immobilization, and 12-hr-immobilization groups were 5.54 ± 1.16 cm, 4.41 ± 0.81 cm, and 4.50 ± 0.79 cm, respectively). Similar drifts have been previously described in the same no-visual-feedback condition (Brown, Rosenbaum, & Sainburg, 2003a). At test, the distribution of movement onsets at the end of the block was significantly different from baseline only in the 12-hr immobilization group, F(2, 18) = 9.76, p < .02 (see Figure 2A).

FIGURE 2.

The drift of movement starting point across trials differed in the three groups. (A) Mean onset point at end of baseline (open circles) and test (closed circles). Ellipses indicate standard errors. Notice the difference between baseline and test in the 12-hr immobilization (IMM) group. The distribution of onset points differed significantly only in that group. (B) Graphical representation of joint displacement. Dotted lines represent starting limb configuration, thin and thick black lines represent the average of limb configurations at end of baseline and end of test, respectively. (C) Difference between test and baseline in each cycle of six movements for movement onset point and for elbow and shoulder angles. That difference, which represents the change of the drift between test and baseline, was significantly higher in the 12-hr IMM group than in the other groups.

Shoulder and elbow configuration drifted accordingly (see Figures 2B and 2C). At baseline, all participants' shoulder angle at movement onset increased approximately 4−5° (see Figure 2B). No consistent changes in elbow angle were found in the three groups. Therefore, the drifts of movement onset were mostly related to changes in shoulder angle (see Figure 2B). During test, changes of shoulder and elbow angles were similar to baseline in no-immobilization and 6-hr-immobilization groups (see Figures 2B and 2C). In the 12-hr-immobilization group, the arm at movement onset instead drifted to a slightly more flexed position because of an increase in shoulder angle of approximately 7° (see Figure 2B). In addition, the forearm assumed a more extended position because participants significantly decreased the elbow angle approximately 10° (see Figure 2B). The differences in elbow and shoulder angles between baseline and test are shown in Figure 2C. ANOVAs showed a significant effect of group: For movement onset point, elbow angle, and shoulder angle, Fs(2, 18) = 13.53, 51.24, and 35.55, respectively, all ps < .00001. Post hoc tests revealed that the difference between test and baseline was higher in the 12-hr-immobilization group than in the 6-hr-immobilization and no-immobilization groups, p < .0001 in each comparison.

At test, all participants' mean movement extent, mean peak velocity, and mean directional error did not change significantly from baseline (see Table 1). However, visual inspection of hand pathways revealed that the shape of the trajectories was altered after 12 hr of immobilization (see Figure 3A): Movement reversals were less sharp, out-and-back strokes did not overlap, and thus the area enclosed in the hand path was increased. Those trajectory changes were not present in either the no-immobilization group or the 6-hr-immobilization group. To capture those alterations, we computed the normalized hand-path area (see Method). Both mean and variability (SD) of hand-path area increased at test only in the 12-hr-immobilization group by 27.9% (± 6.4%) and 32.5% (± 6.6%), respectively (see Figures 3B and 3C). Controls showed a minor, but not significant, decrease at test, whereas the 6-hr-immobilization group showed no significant changes (see Figures 3B and 3C). ANOVA disclosed a significant Group × Block interaction for mean normalized area, F(2, 54) = 7.50, p = .001, and post hoc tests revealed a significant difference for the 12-hr-immobilization group between test and baseline, p = .004 (see Figure 3B). ANOVA on variability of normalized hand-path area disclosed a significant interaction between block and group, F(2, 18) = 7.76, p = .0037, and post hoc tests revealed a significant increase at test only after 12 hr of immobilization, p = .004 (see Figure 3C).

FIGURE 3.

Hand-path changes at test. (A) Representative trajectories to 45° target for a representative participant from each group at test. (B) Mean normalized hand-path area for the three groups at baseline (open bars) and at test (closed bar). At test, mean normalized area slightly decreased in controls and increased slightly after 6 hr of immobilization (IMM 6 group) and consistently after 12 hr (IMM 12 group). (C) Variability of normalized hand-path area across the three target directions (standard deviation of the mean). That parameter significantly increased at test only after 12 hr of immobilization. (D) Variability of interjoint timing (IJT) increased after 12 hr of immobilization.

In a previous study, Sainburg et al. (1995) found that the increases in hand-path area and its variability result from defects in interjoint coordination. We thus computed IJT (i.e., the temporal difference between elbow and shoulder reversals in each movement; Sainburg et al.). To execute a sharp reversal with overlapping strokes, the performer should ideally coordinate elbow and shoulder perfectly, with IJT tending to zero. When interjoint coordination is impaired, movements become less accurate and round and have greater IJT variability. IJT is meaningful only for movements that require both elbow and shoulder motions. For that reason, we focused our analysis on movements in the 45° direction (i.e., actions that require the movement of two joints; Sainburg et al.). Variability (SD) of IJT increased at test significantly only in the 12-hr-immobilization group, whereas it did not change in the other two groups (see Figure 3D): Group × Block, F(2, 2) = 4.90, p = .01; post hoc p < .005. Post hoc tests also revealed that IJT variability at test was significantly higher in the 12-hr-immobilization group than in the other groups, p < .01 for each comparison. In the 12-hr-immobilization group, mean IJT also increased at test, although it did not reach significance: baseline = 3.61 ± 0.94 ms, test = 6.81 ± 1.92 ms, p = .07. We found no differences in the other two groups at test.

The increase of normalized area and IJT variability changes were positively correlated, r = .6, p = .007, suggesting that the changes in hand path induced by immobilization were partially the result of altered interjoint coordination.

We then ascertained whether changes in hand-path areas depended on the drift of movement onset. There was no correlation between hand-path area increases and drift changes, r = .304, p = .18. To confirm that conclusion, we asked the second group of participants to perform reaching movements, with visual cursor feedback provided once every three movements so that they could realign movement onset with visible starting location (see Method). In those testing conditions, drift of the onset position did not occur at baseline or at test in the control and immobilized groups (see Table 2): For group, F(1, 23) = 1.36, p = .26; for block, F(1, 23) = 0.06, p = .81; and for Group × Block, F(1, 1) = 2.3, p = .14. However, as in Experiment 1A, after 12-hr immobilization, mean and variability of normalized hand-path area significantly increased by 26.9% ± 5.7% and 37.8% ± 4.2%, respectively, p < .05 (see Table 2). With regard to mean normalized area, F(1, 23) = 2.1, p = .16, for group; F(1, 23) = 4.4, p < .05, for block; and F(1, 1) = 4.2, p < .05, for the interaction of Group × Block. With regard to area variability, F(1, 23) = 6.75, p = .016, for group; F(1, 23) = 5.2, p = .03, for block; and F(1, 1) = 10.3, p = .004, for the Group × Block interaction. In the no-immobilization group, variability at test decreased slightly, although not significantly, by 13.7% ± 6.9% (see Table 2).

TABLE 2.

Mean (± Standard Error) Onset Drift, Normalized Area, and Normalized Area Variability of Controls and 12-hr Immobilized (IMM) Participants

	Controls		12-hr IMM
Condition	M	SE	M	SE
Onset drift (cm)
Baseline	0.131	0.070	0.041	0.016
Test	0.065	0.054	0.068	0.025
Mean normalized area
Baseline	0.053	0.004	0.051	0.002
Test	0.050	0.005	0.073	0.007*
Normalized area variability
Baseline	0.057	0.003	0.053	0.002
Test	0.051	0.003	0.092	0.009*

^*p < .05.

Discussion

We showed in this study that 12 hr of immobilization produce alterations in trajectory formation and interjoint coordination that qualitatively resemble those previously described in patients with proprioceptive deficits (Sainburg et al., 1995). In addition, we found that when participants performed movements without visual feedback, the point of movement onset drifted more considerably after immobilization. Most interesting, altered trajectory formation also occurred when onset drift was not present in trial blocks with visual feedback at movement onset. Because participants involved in this study have normal proprioception, it is likely that the alterations we found resulted from changes in proprioceptive memory of limb internal models. Immobilization probably induces those changes at the cortical level.

Hand-Path and Interjoint Coordination Abnormalities

Ghez et al. (1995), Gordon et al. (1995), and Sainburg et al. (1995) previously showed that in deafferented patients, trajectory accuracy was severely degraded, with important initial directional errors, abnormal reversals, round shape, and increased area. The errors resulted from uncoupling of joint motions, because the activity of agonist and antagonist muscles was not properly coordinated with the interaction torques that developed between the limb segments during the course of motion (Sainburg et al.).

The changes induced by immobilization were in fact smaller than those of deafferented participants: Alterations of hand-path shape were less prominent, worsening of interjoint timing was modest, and there was no significant change in directional error. Those disparities may partially stem from differences in the present two experimental designs, but they more likely stem from differential alterations of proprioceptive functions. The central nervous system must use proprioceptive information to adjust for the interaction torques that develop during movement. However, because of delays in neural transmission and muscle contraction, online feedback is not sufficient to prevent interaction torques from producing significant errors: Because torques are large when movement is fast and position is changing rapidly, corrective responses will occur when they are no longer appropriate. Therefore, it is likely that feed-forward mechanisms, which are based on internal models or memories of limb dynamics, must assume a prime role in controlling intersegmental dynamics. We found that 12 hr of immobilization reduce the activation of proprioceptive afferents, mostly Ia, and may result in changes of sensorimotor memories. Those changes, in the presence of intact proprioception, which can provide reliable feedback signals, may induce errors that are qualitatively similar to, but of lesser extent than, those of deafferented patients, whose feedback and feedforward mechanisms are altered.

Further support for the hypothesis that alteration of proprioceptive memory is involved in trajectory changes comes from changes in median SEPs (Huber et al., 2006). After 12 hr of immobilization, the amplitude of the P45 component—which represents the proprioceptive information processing within the sensorimotor areas (Allison, McCarthy, & Wood, 1992)—was significantly less than baseline values. The P45 amplitude decrement was highly correlated with the changes in hand paths, suggesting that the trajectory alterations express modifications of proprioceptive cortical representation induced by immobilization.

Interjoint timing and hand-path area did not change after 6-hr immobilization. However, in the control group, which did not go through immobilization, we found a modest decrease in movement area at test. Such improvement probably resulted from consolidation or was practice related, as demonstrated in various learning studies on implicit motor skills (Brashers-Krug, Shadmehr, & Bizzi, 1996; Krakauer, Ghez, & Ghilardi, 2005). Thus, 6 hr of limb immobilization may have prevented the improvement by causing the motor system to modify the internal representation of limb dynamics.

Drift of Movement Onset Increases With Immobilization

Immobilization also induced an increase in the drift of movement onset that is present when vision is withheld during a trial block. Indeed, the results of several studies have shown that in the absence of visual feedback, the onset point of successive out-and-back movements progressively drifts from the initial starting point (Brown et al., 2003a; Brown, Rosenbaum, & Sainburg, 2003b). Consequently, limb configuration at movement onset also changes during successive movements, whereas direction, extent, and other movement characteristics remain remarkably invariant. After 12 hr of immobilization, we found that the displacement of movement onset from the original starting point increased significantly. What could cause that increase in onset drift? Because proprioception is a reliable source of information for movement control (Brown et al., 2003a, 2003b), vision is necessary for position control. Sainburg and colleagues have recently proposed that there are separate controllers for movement and position (Brown et al., 2003a) that depend either on proprioceptive or visual input, respectively. According to the separate controllers hypothesis, the immobilization-induced drift enhancement is probably not the result of the movement control system. If it were, extent and direction of the movement would also be significantly altered. We actually found that extent and direction were preserved. Thus, the most likely scenario is that the function of the position controller—which cannot properly transform the intrinsic information about limb configuration in the absence of visual information—is further compromised by immobilization-induced changes of internal models. The resulting positional errors were not very large and, very plausibly, the proprioceptive system could not compensate for the errors because it was unable to detect them. Our data also suggest that the alterations in hand-path area are not linked to the drift of movement onsets. First, we in fact found that changes of hand-path area did not significantly predict the changes of the onset position. Second, we still found hand-path alterations of the same magnitude when we allowed vision of the cursor to prevent the onset drift.

Cortical Plasticity and Immobilization

The results of SEP studies suggest that the effects of limb immobilization on trajectory formation are most likely related to plastic changes in the sensory cortex (Huber et al., 2006). Motor maps or cortical excitability also change after several weeks of immobilization (Kaneko et al., 2003; Zanette, Manganotti, Fiaschi, & Tamburin, 2004) and even after shorter periods (Facchini et al., 2002). Huber et al. found that 12 hr of immobilization led to selective decrease in MEP amplitude. It is most interesting that, in subsequent sleep, Huber et al. also found a decrease of slow wave activity over the sensorimotor areas contralateral to the immobilized limb. Slow wave activity, which is probably the expression of the synaptic remodeling that takes place during sleep (Tononi & Cirelli, 2003), is a local phenomenon that is proportional to specific learning (i.e., synaptic potentiation; Huber, Ghilardi, Massimini, & Tononi, 2004) or unlearning (i.e., synaptic depression; Huber et al., 2006) that has occurred before sleep. The selective reduction of slow wave activity in sensory and motor areas following 12 hr of immobilization suggests that cortical synaptic depression had occurred.

Researchers must consider whether changes in muscle structure or motor units are responsible for the immobilization-induced changes. That seems very unlikely. Results of recent studies, including ours, suggest that short periods of immobilization do not affect the functions of peripheral nerve and muscle (Facchini et al., 2002; Huber et al., 2006). Seki et al. (2001a, 2001b) also showed that muscle properties in humans undergo reversible changes with a transition from slow fiber type to fast fiber type. However, that transition occurs only after several weeks of immobilization. In animal studies, researchers have similarly shown that cellular levels of several messenger RNAs involved in aspects of metabolism and muscular structure are perturbed by 12 hr of decreased activity (Bey et al., 2003), but those processes take several weeks to produce muscle fiber transition. Therefore, the changes in motor performance occur too early to be ascribed to changes in muscle structure.

The present findings may have important therapeutic implications because health care practitioners use limb immobilization as a rehabilitation tool to address some neurological problems. Constraint-induced therapy, which is based on immobilization of the unaffected arm and extensive training of the affected one, is used in the recovery of hemiplegic patients after stroke (Liepert et al., 1998; Taub & Uswatt, 2006). The therapy results in enlargement of the motor map of the affected muscle and restriction of the map of the immobilized one (Liepert et al.). Neurologists have used immobilization in occupational dystonia, a hyperplasticity disorder, to shrink the corresponding motor areas and reduce the dystonic features (Priori et al., 2001). Following 4−5 weeks of limb immobilization, patients improved, and the improvement persisted for weeks. The present results suggest that the reported changes are not the only effects induced by therapeutic immobilization and that motor performance may also be selectively affected.

Biographical Notes

Clara Moisello is a doctoral student in bioengineering. Motor control and motor learning in humans and in neurological diseases are her main research interests.

Marco Bove teaches human physiology. His main research interests are motor control and rehabilitaton.

Reto Huber's main research interests are sleep and plasticity in humans and during development.

Giovanni Abbruzzese teaches neurology. His main research interests include control of voluntary movement and posture, pathophysiology of movement disorders, and rehabilitation.

Fortunato Battaglia teaches neuroscience and studies plasticity in animals.

Giulio Tononi's main research interests are consciousness and sleep. He has studied sleep by using multiple technologies and has formulated a comprehensive theory on the function of sleep: the synaptic homeostasis theory.

M. Felice Ghilardi teaches neuroscience and neurospychiatry. Her main research interests include motor control, motor learning, and cortical plasticity in humans and the pathophysiology of movement disorders.