Several studies have shown a cortico-spinal facilitation during motor imagery. This facilitation effect is weaker when the actual hand posture is incompatible with the imagined movement. To determine whether the source of this interference effect arises from online proprioceptive information, we examined transcranial magnetic stimulation (TMS)-induced motor-evoked potentials during motor imagery in the deafferented subject G.L. The patient and 7 control subjects were asked to close their eyes and imagine joining the tips of the thumb and the little finger while maintaining a hand posture compatible or incompatible with the imagined movement. Contrary to control subjects' performance, G.L.'s results show that the facilitation observed during motor imagery was independent of the hand posture. To examine how vision of the hand interacts with the imagery process, G.L. and control subjects performed the same task with the eyes open. Like control subjects, when G.L. looked at her hand, a greater facilitation was observed when her hand posture was compatible with the imagined movement than when it was incompatible. These results suggest that in the absence of proprioception, vision may facilitate or inhibit motor representations and support the idea that limb position in the brain is organized around multisensory representations.
Motor imagery may be defined as a dynamic state during which representations of a given motor act are internally rehearsed in working memory without any overt motor output (Decety 1996; Sirigu et al. 1996). Many studies have demonstrated that motor imagery can modulate the excitability of the primary motor cortex in an effector specific manner (Yahagi et al. 1996; Kasai et al. 1997; Kiers et al. 1997; Yahagi and Kasai 1998; Fadiga et al. 1999; Hashimoto and Rothwell 1999; Rossini et al. 1999; Stinear and Byblow 2003, 2004). In recent studies, where subjects had to imagine a finger movement, it has been shown that this motor facilitation effect is posture dependent, that is, strong when the hand is kept in a position compatible with the imagined movement, but significantly weaker when the posture is incompatible with the imagined movement (Vargas et al. 2004; Fourkas et al. 2006). This posture compatibility effect is in accordance with previous findings using similar tasks, which demonstrated that incompatible postural signals affect motor imagery at the time response level (Parsons 1994; Sirigu and Duhamel 2001).
It is still unclear, however, how posture interacts with the motor imagery process. A possible hypothesis is that the conflict arises from peripheral information about the actual hand posture. Alternatively, the conflict could be generated centrally, arising from efferent information about postural motor commands.
To address this question, we used the same paradigm as that used by Vargas et al. (2004) in a patient suffering from a complete sensory deafferention. We measured G.L.'s cortico-spinal excitability during mental simulation of a simple motor task while she held with the eyes closed her dominant hand in 2 different postures: compatible or incompatible with the imagined movement. Because no proprioceptive feedback is available to G.L., a purely central effect would produce interference between posture and motor imagery thus resulting in reduced cortico-spinal facilitation when the posture is incompatible with motor imagery. Alternatively, if the conflict arises from peripheral information we expect no difference on the level of cortico-spinal facilitation between the 2 posture conditions. To further investigate whether visual information about hand posture interacts with the motor imagery process, we then asked G.L. to perform the same task but this time looking at her hand. Because Vargas and colleagues only examined the effects of posture on motor imagery with the eyes closed, in this study we tested 7 additional controls subjects in both eyes-opened and eyes-closed conditions.
Materials and Methods
G.L. is a 56-year-old right-handed female who sustained an episode of extensive sensory polyneuropathy occurring 25 years ago consecutive with a first episode of acute polyneuropathy with a complete paralysis 4 years before (a diagnosis of Guillain-Barré was made at that time). She suffered a permanent, specific loss of the large peripheral myelinated sensory fibers, as confirmed by a biopsy examination (Cooke et al. 1985; Forget 1986; Forget and Lamarre 1995; for more details about G.L. see http://deafferented.apinc.org). Her clinical condition is characterized by the complete loss of light and crude touch, vibration perception, kinesthesia, and position sense in all 4 limbs, as well as in the trunk, neck, and face below the nose. The motor pathways are not affected, with normal motor nerve conduction velocities and an intramuscular EMG study on the arm muscles yielding no sign of abnormality (Forget 1986; Forget and Lamarre 1995).
Seven nonage-matched female controls subjects without any antecedent of neurologic or orthopedic impairment affecting upper limbs were also tested. They were all right-handed females aged between 23 and 31 years old. The study protocol was approved by the Local Ethical Committee (Centre Léon Bérard, Lyon, France). All subjects provided written consent to participate in the study.
It is possible to question the ability of a deafferented subject to perform motor imagery, which is often associated with kinesthetic sensations. Therefore, behavioral testing was performed only in the deafferented subject G.L. to verify that her motor imagery capability was intact. This was done by examining the temporal relationship between execution and imagination using a hand motor task. A temporal congruence between imagined and executed movements has already been demonstrated extensively in normal subjects under a large variety of conditions (Decety and Michel 1989; Parsons 1994; Sirigu et al. 1996; Decety and Jeannerod 1996; Cerritelli et al. 2000).
The subject was asked to imagine joining the tips of the thumb and the little finger (starting with the hand fully opened) while keeping a posture compatible or incompatible with the imagined movement (compatible posture: the little finger, the index and the thumb extended, the remaining fingers flexed; incompatible posture: the index finger and the thumb extended, the 3 other fingers flexed) (see Fig. 1). Prior to the transcranial magnetic stimulation (TMS) testing, the subject was instructed to place the hand in either 1) the posture compatible or 2) the posture incompatible, and then to imagine the thumb-to-little finger opposition movement, repeating opposition of the thumb to the finger from 1 to 5 times (3 trials were performed for each number of repetitions, making a total of 15 trials, and the order of presentation of the trials was randomized). Movement duration was defined as the time between the onset of the auditory go signal and the subject's verbal report that she had completed the task. The subject was then asked to evaluate, on a scale from 1 to 5, the degree of comfort and the ease with which she imagined the movement while maintaining each posture. The subject was also asked to physically execute the movement from 1 to 5 times (as for the imagined condition, 3 trials were performed for each number of repetitions).
Single-pulse TMS was applied over the left primary motor cortex (M1) using a Magstim 200® stimulator with a 70-mm figure-of-8 coil. Precise localization of the TMS coil relative to the brain was achieved by acquiring an anatomical magnetic resonance image (MRI) of the subject's brain and using the image to guide positioning of the coil in real time (Noirhomme et al. 2004). The coregistration of the subject's MRI with the actual position of her head was achieved by measuring the 3-D location of 200 points on the scalp with an electromagnetic position sensor (Polhemus Isotrack II®). Motor-evoked potentials were recorded from the Opponens pollicis (OP) muscle using Ag–AgCl surface electrodes.
The optimal site for stimulating the OP (hotspot) and the motor threshold (intensity at which the TMS stimulation elicited motor evoked potentials (MEP) larger than 50 μV on 50% of the trials) were determined prior to the experimental session (Fig. 2). During the experiment, stimulation was performed at 120% of motor threshold. The use of imaged-guided TMS allowed a precise placement of the coil over the hotspot throughout the experiment (see Fig. 2). To avoid order-related biases each of the 8 conditions (determined by the combination of 1) eyes open/closed; 2) position compatible/incompatible; and 3) rest/motor imagery) were split into 2 blocks of 8 trials making a total of 16 trials by condition and 128 trials in the experiment. The blocks were presented in a randomized order. When the task was performed without vision, the subject adopted the required position before each trial under visual control, and closed her eyes when instructed to get ready, approximately 0.5 s before the “Go” signal. For each trial, the hand posture was carefully monitored by one of the experimenters as we expected that it would be difficult for the deafferented subject to maintain the correct position across time without the assistance of visual feedback. We observed, however, that the trial length (less than 2 s) was shorter than the time over which the patient's hand posture degraded (approximately 5 s).
In order to define a time window for the TMS pulse application, the interval between an auditory go signal and the maximal electromyographic (EMG) activity in the OP muscle was measured while subjects executed the movement. The interval between the go signal and the pulse application was different for each subject and was based upon the EMG measurements and the average duration of executed and imagined movements. During the TMS testing, for each stimulation pulse, the subject always performed a single repetition of the imagined movement (thumb-to-little opposition) and the TMS timing (delay between the go signal and pulse application) was constant and aimed at the EMG peak during the execution.
EMG signals were amplified and band pass filtered (20–1000 Hz, Neurolog instruments, Digitimer Ltd, Letchworth, Garden City, UK). The signal was then digitized at a sampling rate of 2000 Hz (CED 1401 interface, CED Ltd, Cambridge, UK) and stored on an IBM-PC computer for off-line analysis. A Matlab (Mathworks) script was used to measure MEP latency, MEP peak-to-peak amplitude, and prepulse area (over a 50-ms window) from EMG recordings.
Because G.L. has no afferent information about muscle state, it was difficult for her to keep her muscles fully relaxed for a long period. Therefore, we paid particular attention to the amount of EMG activity in OP prior to the TMS pulses. Trials on which EMG activity was observed were rejected online. Moreover, t-tests were performed to verify whether background EMG activity was affected by task (rest/imagery), posture (compatible/incompatible), or visual feedback (vision of the hand/no vision). The only statistically significant difference was found between the 2 visual conditions, (vision/no vision: t(126) = 3.195, P< 0.01), the tonic activity being 30% higher when G.L. was looking at her hand. We believe that this difference might be explained by minor motor adjustments performed by G.L. under visual guidance (because she habitually monitors her posture visually), as no such difference was found for control subjects. An alternative explanation is that vision of the hand induced motor facilitation thus triggering a general increase of tonus in the whole musculature. Because differences in tonic muscle contraction affect MEP amplitude, we analyzed separately the results of the “Vision” and “No vision” conditions for G.L.
In order to compare G.L.'s results with those of normal subjects, MEPs measured during the motor imagery tasks were normalized against the MEPs measured at rest for the same condition of posture and vision. This provided a percentage of facilitation associated with motor imagery for each condition (for example, % of facilitation for compatible posture with eyes opened = (average MEP during motor imagery [compatible posture, eyes opened] − average MEP at rest [compatible posture, eyes opened])/average MEP at rest [compatible posture, eyes opened]). The effect of the different conditions in control subjects was assessed using a 2-way repeated measures analysis of variance (ANOVA). The results of G.L. were compared with those of the control subjects using 95% Bonferroni-adjusted confidence intervals. Individual results of G.L. were also analyzed using t-tests (with appropriate Bonferroni corrections) to determine if there was a posture effect in each vision condition (as mentioned before, the 2 vision conditions were not compared directly because of differences in baseline EMG between the 2 conditions).
When questioned about the perception she had during motor imagery, G.L. always reported performing the task by seeing her hand in action instead of experiencing a kinesthetic feeling. The inability to use a kinesthetic strategy during motor imagery may be due to G.L.'s long lasting deafferentation. Nevertheless, she asserted that she was really mentally “performing” the movement, and not just “looking” at it as a passive agent. G.L. reported imagining the movement easily regardless of the experimental conditions (she always attributed a score of 5 on a 1–5 scale for the ease of imagination). The time required to physically or mentally execute 1–5 repetitions of the task under the different experimental conditions is presented in Figure 3. The time required to physically execute the task significantly decreased with the eyes opened as compared with the eyes closed condition (F1,27 = 62.71, P < 0.001), although she was well able to perform the task with the eyes closed (she was allowed to see the starting posture of her hand before each trial). She commented that without vision she had to “push a little more at the end of the movement just to be sure that her thumb reached her little finger,” which might explain the additional time she needed in this condition. She also took less time to imagine the movement when performing the task with eyes opened (F1,55 = 6.47, P = 0.014), but there was no significant effect of hand posture. Moreover, the time required to imagine was scaled correctly as a function of the number of repetitions and was stable across trials, as illustrated by the small standard deviations.
The percentage of imagery-induced facilitation observed for each condition of vision and posture in control subjects and in G.L. is illustrated in Figure 4.
In control subjects, a 2-way ANOVA showed a significant effect of posture, indicating more imagery-induced facilitation when the hand was in a posture compatible with the imagined movement (F1,6 = 8.998, P = 0.024), but there was no effect of vision and no interaction between the 2 factors.
The only difference between G.L. and control subjects' performance was observed in the eyes closed/compatible posture condition where the large facilitation found in control subjects was not found in G.L. (95% Bonferroni-adjusted confidence interval for control subjects was [49.6–264.7%], whereas G.L. facilitation was of 28.9%). This difference can be attributed to the fact that when G.L.'s eyes were closed there was no effect of posture on the imagery-induced facilitation. Whereas in control subjects the posture affected facilitation regardless of whether the eyes were open or closed, in G.L. this effect was present only when she was allowed to see her hand. In this condition, G.L. showed a greater facilitation in the compatible than in the incompatible posture (t(30)= 3.349, P = 0.002; significance threshold with Bonferroni correction = 0.025).
The results obtained in the “eyes closed” condition show that, in the case of G.L., information arising from the motor command alone may influence but is not sufficient to generate an interaction between the actual hand posture and the motor imagery process. The results from G.L. support the hypothesis that afferent feedback may be responsible for triggering the interaction between actual body posture and motor imagery previously observed in normal subjects (Vargas et al. 2004). This does not completely rule out the contribution of efferent information, because one can argue that efferent signaling in G.L. might be altered as a consequence of her disease or of the long-term adaptation to her condition. Nevertheless, the results of the present study suggest that afferent feedback plays a crucial role in modulating postural information and central motor imagery processes.
It is possible, however, to question these results on the basis of the ability of a deafferented subject to perform motor imagery. Motor imagery is generally considered a process in which the subject feels himself executing a movement while inhibiting the motor output. Normal controls usually report kinesthetic sensations associated with movement imagination (Sirigu et al. 1996), whereas G.L. referred to a visual experience of her hand in action. The observation of an increase in cortico-spinal excitability during imagery as compared with rest suggests that G.L. can perform motor simulation, although the associated sensation is visual rather than kinesthetic. This view is further supported by functional MRI (fMRI) studies showing that in patients with deafferentation (due to amputation or spinal cord injury), the vividness of motor imagery is correlated with the level of activation in the primary motor cortex (Lotze et al. 2001; Alkadhi et al. 2004).
The results obtained in the “eyes opened” condition demonstrate that visual feedback about posture may interact with central motor processes, thereby modulating M1 cortico-spinal excitability. The fact that a posture-dependent modulation of motor excitability was observed when G.L. had her eyes opened further supports the idea that G.L. accomplished the task by actually imagining performing a movement. This is particularly convincing in the light of recent results showing that in control subjects there is an effect of posture on the imagery-induced facilitation during kinesthetic imagery, but not during purely visual imagery (Fourkas et al. 2006). Moreover, the amplitude of the facilitation (vs. rest) observed during motor imagery with eyes opened was in the same range as that observed in control subjects in the same condition.
Several studies have shown that observation of a movement can affect cortico-spinal excitability (Fadiga et al. 1995; Gangitano et al. 2001; Maeda et al. 2002; Aziz-Zadeh et al. 2002; Patuzzo et al. 2003; Stefan et al. 2005) and motor performance (Brass et al. 2001; Craighero et al. 2002; Kilner et al. 2003). Interestingly, M1 excitability can be influenced by interactions between visual and motor activity. Garry et al. (2005) have shown that a unilateral hand movement seen in a mirror (thus giving the visual impression that the contralateral hand is moving) increases excitability in M1 ipisilateral to the movement as compared with a condition where subjects are doing the same movement without this virtual visual feedback. Our results provide further evidence for a close linkage between the visual and the motor systems, even under conditions in which no movement occurs.
The finding that hand posture modulates imagery-induced motor facilitation in control subjects with intact proprioceptive feedback but without vision of the hand, and in a patient with visual feedback but no proprioceptive information supports the idea that limb position in the brain is organized around multisensory representations. Other evidence of multisensory representations of posture comes from studies in nonhuman primates. Graziano (1999, Graziano et al. 2000) has shown that the position of the arm in space is represented in the parietal and the premotor cortex by means of a convergence between visual and proprioceptive information onto the same neurons. These neurons respond in the same fashion to the felt position of the arm when it is obscured from view, and to the seen position of a false arm. Recent fMRI data suggest that cortical areas involved in the multisensory representation of limb position in humans are similar to those reported in nonhuman primates (Lloyd et al. 2003; Ehrsson et al. 2004).
An important issue raised by our results is why the actual hand posture modulates M1 facilitation elicited by motor imagery. It has been shown that at-rest changes in upper limb or hand posture lead to a modulation of the MEPs induced by TMS, even for muscles that are distal to the joint where the postural change occurs, and that these at-rest changes can also lead to a modification in the direction of an evoked movement (Wassermann et al. 1998; Shimura and Kasai 2002; Ginanneschi et al. 2005; Ginanneschi et al. 2006). The fact that a postural change generates excitability variations involving multiple muscles in M1 is supported by the results of studies in nonhuman primates showing that M1 microstimulation can elicit complex movements that cause the relevant joints to move into a specific final posture, regardless of the starting posture (Graziano et al. 2002). It follows then, that different directions of movement can be obtained from the same stimulation site depending upon the starting posture. These results suggest that the activity of the motor system shifts from the actual posture to a desired posture. During a motor imagery task, however, when there is a discrepancy between the actual hand posture and the imagined hand posture, 2 alternative motor strategies compete to attain the desired position, one for motor execution (starting from actual posture) and one for motor imagery (starting from imagined posture). Because there is an overlap of brain networks activated during movement imagery and execution (Decety et al. 1994; Lotze et al. 1999; Gerardin et al. 2000; Stippich et al. 2002) these 2 alternative strategies might compete with each other, resulting in a weaker facilitation of M1 excitability when there is a competition between actual and imagined posture. Conversely, seeing a posture which is congruent with the imagined movement can lead to a summation of the facilitation induced by motor imagery and by visual feedback. This idea is supported by recent studies showing that the simple observation of static snapshots of hands suggesting that a potential action can induce an increase in cortico-spinal excitability (Urgesi et al. 2006). Moreover, observing the end position of a movement is sufficient to induce behavioral motor priming effects (Stürmer et al. 2000).
If hand posture has a multisensory cortical representation, then it follows that the interaction effect between posture and movement simulation may be produced by proprioceptive or visual information. Indeed the results we obtained in G.L. to whom only visual feedback is available provide strong support for this hypothesis. Interestingly, no effect of the condition of vision, nor interaction between vision and posture, was found in control subjects, indicating that there is no additive effect of these 2 sources of sensory feedback (i.e., the interaction effect is not larger when control subjects have access to proprioceptive and visual feedback as compared with proprioceptive feedback only). It has been suggested that such interaction might also result from a purely central effect (Vargas et al. 2004), that is, that efferent information about posture could suffice to modulate motor imagery process. Although the present results do not rule out the contribution of efferent information, the results from G.L. suggest that afferent feedback plays a crucial role in the interaction between posture and central motor imagery processes.
Authors are grateful to M.C. Hepp-Raymond for providing the MRI of G.L. and to K.T. Reilly and E.C. Rodrigues for their help with data collection. This research was founded by National Science Foundation and Centre National de la Recherche Scientifique to A.S. C.M. is supported by FRSQ (Québec), A.A. by Région Rhône-Alpes, and C.D.V. by Capes-Cofecub. Conflict of Interest: None declared.