Neuropsychologia 71 (2015) 46–51
Contents lists available at ScienceDirect
Neuropsychologia journal homepage: www.elsevier.com/locate/neuropsychologia
Enhancing the mirror illusion with transcranial direct current stimulation Steven A. Jax a,n,1, Diana L. Rosa-Leyra a, H. Branch Coslett b a b
Moss Rehabilitation Research Institute, Elkins Park, PA, USA Department of Neurology, University of Pennsylvania School of Medicine, USA
art ic l e i nf o
a b s t r a c t
Article history: Received 25 November 2014 Received in revised form 3 February 2015 Accepted 17 March 2015 Available online 19 March 2015
Visual feedback has a strong impact on upper-extremity movement production. One compelling example of this phenomena is the mirror illusion (MI), which has been used as a treatment for post-stroke movement deficits (mirror therapy). Previous research indicates that the MI increases primary motor cortex excitability, and this change in excitability is strongly correlated with the mirror’s effects on behavioral performance of neurologically-intact controls. Based on evidence that primary motor cortex excitability can also be increased using transcranial direct current stimulation (tDCS), we tested whether bilateral tDCS to the primary motor cortices (anode right-cathode left and anode left-cathode right) would modify the MI. We measured the MI using a previously-developed task in which participants make reaching movements with the unseen arm behind a mirror while viewing the reflection of the other arm. When an offset in the positions of the two limbs relative to the mirror is introduced, reaching errors of the unseen arm are biased by the reflected arm’s position. We found that active tDCS in the anode right-cathode left montage increased the magnitude of the MI relative to sham tDCS and anode left-cathode right tDCS. We take these data as a promising indication that tDCS could improve the effect of mirror therapy in patients with hemiparesis. & 2015 Elsevier Ltd. All rights reserved.
Keywords: Mirror therapy Mirror feedback Reaching Transcranial direct current stimulation
1. Introduction The influence of visual feedback on upper extremity movement production is well documented (Desmurget et al., 1998). One compelling application of this finding comes from mirror therapy (MT) as it is used to treat post-stroke hemiparesis. Originally developed as a treatment for phantom limb pain (Ramachandran and Altschuler, 2009), MT involves the patient being seated in front of a vertically-oriented mirror (Fig. 1). The patient places her unimpaired arm on the reflective side of the mirror and her impaired arm behind the mirror so that it is hidden. When the mirror is placed midway between the two limbs, movements of the unimpaired limb (viewed in the mirror) appear in the same location as the impaired limb. Thus, the MT setup creates a compelling illusion in which movements of the impaired arm behind the mirror appear to be made as effectively as the unimpaired arm. Although previous clinical research indicates that MT is efficacious (Thieme et al., 2013) the underlying neural changes associated with MT are poorly understood. Because MT-like effects can also be observed in neurologically-intact controls (which we will n
Corresponding author: E-mail address: [email protected] (S.A. Jax). 1 50 Township Line Rd., Elkins Park, PA, 19027, USA.
http://dx.doi.org/10.1016/j.neuropsychologia.2015.03.017 0028-3932/& 2015 Elsevier Ltd. All rights reserved.
term the “mirror illusion,” or MI, to differentiate it from the therapeutic use of the mirror setup), several studies have examined the possible neural mechanisms of the MI. Many of these studies, as well as studies of stroke survivors, suggest that mirror feedback modifies functioning within the primary motor cortex of the hemisphere that controls the arm behind the mirror (the lesioned hemisphere in stroke survivors). In controls, the MI increases primary motor cortex excitability (Garry, et al., 2005) and this change in excitability is strongly correlated with the mirror’s influence on movements of the arm behind the mirror (Nojima et al., 2012). These changes in excitability may come about via intra- or inter-hemispheric neuroplasticity (Hamzei et al., 2012; Nojima et al., 2012). In stroke survivors, effective connectivity between the primary motor cortex and the somatosensory cortex in the lesioned hemisphere is increased under single-session MT conditions (Saleh et al., 2013) and completing 6 weeks of MT increases BOLD activation of the lesioned hemisphere’s primary motor cortex relative to the intact hemisphere’s primary motor cortex (Michielsen et al., 2011). While many studies indicate that the mirror feedback in the MI and during MT primarily changes neural functioning within the hemisphere controlling the hand behind the mirror, conflicting findings have been reported. For example, increased reliance on regions within the hemisphere controlling the visible hand have
S.A. Jax et al. / Neuropsychologia 71 (2015) 46–51
47
“Predictions” section below. We used this task as an index of the effect that applying bilateral primary motor cortex tDCS had on the MI. Previous research indicates that anodal tDCS increases primary motor cortex excitability whereas cathodal tDCS decreases excitability (Nitsche and Paulus, 2000). If tDCS’ effects on primary motor cortex excitability are similar to those reported from the MI (Garry et al., 2005; Nojima et al., 2012) then applying tDCS during the task should affect the MI. Predictions about the effect of tDCS depend on the presumed site of the MI effect and the tDCS montage. If one assumes that the MI results in increased excitability within the hemisphere primarily controlling the hand behind the mirror (the right hemisphere in the present study), then applying anodal stimulation to the right hemisphere and cathodal stimulation to the left hemisphere should increase the MI. Similarly, reversing the polarity of stimulation (right cathode-left anode) should reduce the MI. In contrast, if one assumes that the MI results in increased excitability within the hemisphere controlling the hand in front of the mirror (the left hemisphere in the present study), then the MI should be increased in the right cathode-left anode montage and decreased in the right anode-left cathode montage. The present study tested these contrasting predictions.
2. Methods 2.1. Participants Fig. 1. Reaching task setup. The participant’s right hand covers the near (R1) start location, while the dot indicating the R2 start location is visible to the right of the hand. The white box above and to the left of the mirror highlights the two dots that indicated the position of the imagined target locations along the platform, which the participant reached for with the unseen left hand.
Twelve college-aged young adults [mean (s.d.) age 21.3 (2.4) years old; 5 females] completed the study. All participants were recruited through the University of Pennsylvania and received monetary compensation for their participation. 2.2. tDCS stimulation
been reported using TMS (Läppchen et al., 2012) and neuroimaging (Hamzei et al., 2012) methodologies. In addition, all of the studies cited in the previous paragraph except one (Michielsen et al., 2011) involved no movement of the hand behind the mirror. These conditions differ from those typically used during MT, in which patients attempt to move both limbs simultaneously (Dohle et al., 2009; Michielsen et al., 2011). It is unknown whether the neural changes associated with uni- and bi-manual MT/MI are different, but one study in patients suggests that these conditions can lead to different behavioral outcomes (Selles et al., 2014). Thus, to date there is conflicting evidence about the neural mechanisms of MT. Resolving this conflict may have significant therapeutic significance, as understanding these mechanisms could be used to augment MT’s endogenous neural changes with exogenous non-invasive brain stimulation such as transcranial direct current stimulation (tDCS). The present study was designed to provide an initial step towards this goal by testing whether applying tDCS to the two primary motor cortices would affect the MI in neurologically-intact controls. Given our goal of applying the results of this study to post-stroke rehabilitation, we wanted to examine the effects of tDCS on motor performance when bimanual movements were sometimes required. To do so, we utilized a task developed by Holmes and Spence (2005) to elicit the MI. Participants began by making symmetric bilateral movements within the mirror setup. Then, participants made a reaching movement with the unseen left arm (behind the mirror) while viewing the reflection of the stationary right arm. When an offset in the positions of the two limbs relative to the mirror was introduced, the unseen hand’s trajectory was biased as if it were starting nearer the position of the mirror-reflected right hand, an effect not observed when the mirror was covered. A more detailed description of how the task allowed quantification of the MI will be provided in the
Each participant completed three separate single-day testing sessions, each of which employed a different tDCS montage. In all sessions, two 5 7 cm2 saline soaked electrodes were positioned over C3 and C4 of the International 10–20 EEG placement system. Within each session, a Magstim tDCS stimulator provided 1.5 mA of current in one of three montages: (1) right cathode-left anode, (2) left cathode-right anode or (3) sham. In the sham montage the same electrode locations were used (half of participants had right cathode-left anode, half had left cathode-right anode) but with a 30 s ramp up then 30 s ramp down of 1.5 mA simulation to provide tactile sensations similar to the other montages. Stimulation began at the start of each reaching task and continued until the task was complete (approximately 20 min). Each session was separated by at least two weeks to minimize any carry-over effects of both tDCS and the reaching task. 2.3. Reaching task Within each session, participants completed the same reaching task. Participants sat slightly to the right of the vertically-oriented mirror (30 cm 30 cm) so they could see the reflection of their right arm in the mirror (Fig. 1). Vision of the left arm was occluded by the mirror as well as by black fabric draped between the front of the apparatus and the participant’s left shoulder. Each trial began when the experimenter instructed the participant to place his or her right index finger on one of two starting positions indicated by a red and blue dot on the platform (9 and 16 cm to the right of the mirror; approximately 19 cm in front of the participant’s body). Next, the experimenter moved the participant’s left arm to one of the four left hand start positions (9, 16, 23 and 30 cm to the left of the mirror; approximately 19 cm in front of the
48
S.A. Jax et al. / Neuropsychologia 71 (2015) 46–51
participants body). Once in position, participants were instructed to synchronously flex and extend both wrists in time with a 120 bpm metronome for 8 s while gazing at the right arm in the mirror. These wrist movements were included to elicit the mirror illusion in a consistent fashion before each trial, a technique similar to the one used by Holmes and Spence (2005). Next, participants used their left arms to reach forward and touch an imagined target located on the platform directly below either a red or blue dot visible on the frame of the apparatus (highlighted with a white box in Fig. 1). The target to which subjects reached was the same color as the right hand start dot. While participants performed the task, movements of the left index finger were recorded using an Ascension TrakSTAR kinematic recording system sampling at 150 Hz so that endpoint reaching errors could be measured. A sensor was also placed on the right index finger so that tactile sensation was similar across the hands. Each block included 16 trials randomly comprising two repetitions of all combinations of the two right hand starting positions (R1, R2) and four left hand starting positions (L1, L2, L3, L4). Each session included four blocks, two with the mirror visible and two with the mirror completely covered so that the reflection of the right arm was not visible. The order of mirror and covered blocks was counterbalanced within the session using an ABBA order. The status of the session's first block (mirror or covered) was counterbalanced across participants and sessions. Thus, within each session participants completed 64 trials, or four repetitions of the 16 possible combinations of the two mirror conditions (mirror, covered), the two right hand starting positions, and the four left hand starting positions. 2.4. Predictions Fig. 2 illustrates how the experimental method allowed us to
determine the effect that the mirror-reflected right hand had on reaching in the unseen left hand (the MI). Predictions are best illustrated by considering the left hand’s endpoint errors along the left-right axis under two extreme possible outcomes: when the mirror-reflected right hand has no effect on the left hand (left column of Fig. 2), and when the mirror-reflected right hand defines the location of the left hand in left hemispace (middle column). In the first example (A and B), the left hand (left triangle at position L1 in A) starts the trial closer to the mirror than the right hand (right triangle at position R2; note that targets R1 and L1 are equidistant from the mirror, as are R2 and L2). If the position of mirror-reflected right hand has no effect on the left hand’s error, the left hand should move directly to the target (solid arrow in A). In contrast, if the reflected right hand’s position defines the position of the left hand in left hemispace, an accurately planned movement would result in the left hand moving parallel to the mirror directly to the target (dashed arrow in B). However, since the left hand is actually located closer to the mirror than the right hand, executing the same movement parallel to the mirror from the actual starting position would result in an endpoint error that is to the right of the target (solid arrow in B). Thus, if the reflected right hand’s position is taken to be the position of the left hand in left hemispace, endpoint errors will be biased towards the starting position of the left hand. A second example (C and D) illustrates this directional effect. In contrast to the previous example, the left hand (at L3) starts further from the mirror than the right hand (at R2). As in the previous example, if the mirror has no effect the participant will reach directly to the target. However, if the reflected right hand is taken to indicate the position of the left hand in left hemispace, the left hand should again move parallel to the mirror directly to the target (dashed arrow in D). Because the left hand is further from the mirror than the right hand, the endpoint error is biased to the
Fig. 2. Examples of predicted results under the hypothesis that the mirror will not affect left-hand reaching (left column; A, C, E) and under the hypothesis that the reflected right hand’s position will be taken to be the position of the left hand in left hemispace (middle column; B, D, F). In Panels B, D, and F, dashed arrows indicate the proper movement vector if the left hand position was assumed to be in the location of the mirror reflected right hand, and solid arrows indicate the corresponding movement path from the actual starting position if that vector were to be executed. Predictions across all left hand start locations are summarized in Panel G for the R2 start position. See Predictions section in the text for details.
S.A. Jax et al. / Neuropsychologia 71 (2015) 46–51
49
left of the target (solid arrow in D). Like the first example, the leftright endpoint errors are biased towards the starting position of the left hand. The magnitude of this bias is further increased when the offset between the hands is increased (E and F). To summarize the predictions of these two extreme possible outcomes (no effect of mirror and total effect of mirror), one can plot the expected left-right endpoint errors as a function of the four left hand start positions for a single right-hand start position (e.g. R2, as illustrated in this figure). Under the outcome in which the mirror will have no effect, reaching errors will be observed (because reaching is done without visual feedback), but the left-right bias of the errors is predicted to be near zero and be unaffected by changes in the left hand start position. That is, a line connecting the error bias of the four left hand start positions should have a slope near zero (dashed line in G). In contrast, under the outcome that the mirror-reflected position of the right hand will be taken as the position of the left hand in left hemispace (total effect of the mirror), endpoint errors will be biased to the right when the left hand starts closer to the mirror than the right hand (B) and biased to the left when the left hand starts further from the mirror than the right hand (D and F). Therefore, a positive slope in the line connecting the errors of the four left hand start positions would indicate evidence for the MI (solid line in G), with the magnitude of the slope quantifying the magnitude of the mirror’s effect. Based on the results of Holmes and Spence (2005) we predicted that the slopes of the error lines would be greater in the mirror conditions than the covered conditions. In addition, we tested whether slopes in the mirror condition would be affected by tDCS. As described in the Introduction, if the MI results in increased excitability within the hemisphere controlling the hand behind the mirror (the right hemisphere in the present study), then slopes in the mirror condition should increase in the anode right condition and decrease in the anode left condition. In contrast, if one assumes that the MI results in increased excitability within the hemisphere controlling the hand in front of the mirror (the left hemisphere in the present study), then the opposite pattern should be observed. That is, slopes should increase in the anode left montage and decrease in the anode right montage.
3. Results The average left-right endpoint error data for all conditions are shown in Fig. 3. In this figure, two central results can be observed. First, slopes in the error lines for all conditions were steeper in the mirror conditions (solid lines) than the covered conditions (dashed lines). Second, mirror condition slopes of the right-anode condition were steeper than the other two tDCS conditions. To confirm these results using inferential statistics in the most direct and transparent way2, we used regression analysis to compute the best-fitting regression line predicting the left-right endpoint errors (in cm) for the L1, L2, L3, and L4 starting positions with the corresponding distances the left hand was from the 2
An alternative analysis approach that we considered but rejected was analyzing the endpoint error data shown in Fig. 3 with a 4-way ANOVA with the factors tDCS (anode right, anode left, sham), Right hand start location (1,2), Left hand start location (1, 2, 3, 4), and Mirror (mirror, covered). Using this approach, one would need to examine interactions to test the central hypotheses about slopes. For example, one would look for an interaction between the Left hand start location factor and the Mirror factor to show that slopes were steeper in the mirror condition than the covered condition. The limitation of this ANOVA approach is that testing for an interaction is, graphically, only a test to determine if the lines are not completely parallel to one another. Thus, although all differences in slope would lead to an interaction, not all interactions would indicate differences in slope. Therefore, we choose to directly compute slope values and then analyze those slopes using an ANOVA.
Fig. 3. Mean left-right endpoint error data (7 1 s.e.) for all conditions. Separate panels illustrate results from the three tDCS conditions (left anode, right anode, and sham). Within panels, errors are shown separately for the mirror (solid line) and covered (dashed line) conditions. Lines connect errors for the four possible left hand start locations (L1, L2, L3, and L4) separately for the two right hand start locations (R1 or R2).
mirror at the start of the trial (9, 16, 23 and 30 cm, respectively). Note that the value of the slope indicates the proportion of the possible MI effect that was observed because it measures how many centimeters one would expect endpoint error to change with an additional 1 cm shift in left hand start location. Thus, a
50
S.A. Jax et al. / Neuropsychologia 71 (2015) 46–51
Fig. 4. Mean slope values ( 71 s.e.) for the mirror and covered conditions, reported separately for the three tDCS conditions (separate bars) and two right hand start locations (R1 and R2).
slope of 0 indicates 0% of the possible MI effect observed, and a slope of 1 indicates 100% of the possible MI effect. Regressions were completed at the participant level, and for each participant 12 best-fitting lines were computed for each combination of tDCS (anode right, anode left, sham), mirror (mirror, covered), and right hand start position (1, 2). The slopes of these 144 lines (12 participants 12 condition combinations per participant) were then analyzed using a 3 (tDCS: anode right, anode left, (Right hand start: 1,2) 2 (Mirror: mirror, covered) ANOVA, with all factors manipulated within participants. The two sham configurations were grouped because analyses indicated no differences between the montages. The results of the slope analysis ANOVA are shown in Fig. 4. There was a main effect of Mirror, F(1,11) ¼43.72, p o.001, with average slopes being higher in the mirror conditions than the covered conditions (slope of 0.099 vs. 0.033, respectively). There was also a main effect of tDCS, F(2,22) ¼4.68, p¼ .020. Bonferronicorrected follow-up t-tests indicated that average slopes were significantly higher (p¼ .025) in the anode right condition than the sham condition (slope of 0.080 vs. 0.059), and that average slopes in the anode right condition were marginally (p ¼.067) higher in the anode right than the anode left condition (slope of 0.080 vs. 0.058). Most importantly, the Mirror and tDCS factors interacted significantly, F(2,22) ¼ 4.25, p ¼ .027. When the effects of tDCS were analyzed separately for the mirror and covered conditions, there was no effect of tDCS in the covered condition, F(2,22) ¼.056, p ¼.946, whereas this effect was significant in the mirror condition, F(2,22) ¼8.52, p¼ .002. Bonferroni-corrected follow-up t-tests indicated that average slopes in the anode right condition were significantly higher than the anode left condition (slope of 0.126 vs. 0.087, p ¼.037) and the sham condition (slope of 0.126 vs. 0.083, p ¼.008). No effects of Right hand start, nor any interactions between this factor and other factors, were observed (p's 4.37).
4. Discussion In this study we showed for the first time that the mirror illusion could be increased using noninvasive brain stimulation. When concurrent anodal tDCS was delivered to the primary motor cortex controlling the unseen arm behind the mirror (along with cathodal stimulation to the primary motor cortex controlling the visible arm), reaching errors of the unseen arm were more biased by the reflected arm’s position than when sham stimulation or stimulation with the opposite polarity was delivered. This effect of tDCS was not observed in the control condition in which the
mirror was covered. Thus, tDCS specifically affected the way the mirrored visual feedback was utilized by the motor system, and the effects of tDCS we observed cannot be ascribed to general changes in reaching with the left arm. Critical for our goal of applying this result to post-stroke motor rehabilitation, the tDCS montage that increased the MI was the same montage that has been used therapeutically to treat hemiparesis under non-mirrortherapy conditions (Mahmoudi et al., 2011). Thus, this study provides a promising possibility for improved post-stroke rehabilitation by combining MT and tDCS in future studies. In the remainder of the Discussion section, we will discuss (1) the role that proprioception may have played in our results and (2) the implications of our finding for the neural substrates of the mirror illusion and mirror therapy. In the mirror illusion, the visually-derived estimate of the location of the arm behind the mirror (the left arm in the present study) is manipulated. It is important to note that proprioception also plays an important role in limb position estimation (Van Beers et al. 1999). The influence of proprioception to the MI is clearly observed in two results in the present study. First, the magnitude of the MI in the mirror condition was a fraction of the total possible MI, defined as the predicted results if the mirror-reflected right hand was taken to be the location of the left hand in left hemispace (see Fig. 2). Across the tDCS conditions, the slopes of the error bias were between .08 and .14, or 8–14% of the possible MI. These slope were lower than slopes in comparable conditions of Holmes and Spence (2005), which were approximately .30 or 30% of the possible MI. Further replication will be needed to better understand this difference in MI magnitude from two studies using highly similar, but not identical, methods. Nonetheless, the effects of the mirror were consistent, but did not completely “trick” the participant into believing the reflected arm truly indicated the left arm’s location. This incomplete effect is likely due to the proprioceptive signal indicating the left arm’s actual location, which can then be modulated by the visual illusion provided by the mirror. The limited effect of the mirror may have also been due to the relatively large possible asymmetries in the positions of the two arms. Under the most extreme condition, the right hand started 9 cm from the mirror while the left hand started 30 cm from the mirror. The MI would have to quite strong to completely overcome the proprioceptive signals about the hidden arm's position with such an offset. The effects of proprioception were also likely observable in the slightly positive slopes in the covered conditions, which indicated that the right hand’s position affected reaching with the left hand, even when the mirrored visual feedback was not present. This finding may have been caused by the motor system’s tendency to assume proprioceptive symmetry between the limbs, such that the proprioceptively-sensed location of the right arm would provide some influence of the estimate of the left arm’s position. Thus, at least some of the MI may be caused by proprioceptive mirroring. However, the visual input clearly matters given that slopes were at least twice as great in the mirror condition than the covered condition. If proprioceptive symmetry were the sole cause of the MI, results would be predicted to be similar across the mirror and covered conditions. Given the role that proprioception can play in this task, it is logical to ask whether applying tDCS centered on the primary motor cortex may have also stimulated the primary somatosensory cortex, and therefore also affected proprioceptive processing. Given the non-focality of tDCS, this is a possibility. However, the effects of tDCS were only seen in the mirror condition. Thus, it is unlikely that our results were due to changes in the way that all proprioceptive information is processed because such changes would have affected both the mirror and covered conditions similarly. However, it is possible that tDCS affected how visual and
S.A. Jax et al. / Neuropsychologia 71 (2015) 46–51
proprioceptive information were integrated. Future studies with more focal forms of stimulation (such as transcranial magnetic stimulation) could investigate this issue further. Relatedly, even though we applied tDCS to the primary motor cortices, the neural changes responsible for the modulation of the MI need not necessarily emanate solely from the primary motor cortex. Visuallyguided upper extremity movements are controlled by a network of brain regions, and the effects of tDCS are known to extend beyond the regions underneath the electrodes (Brunoni et al., 2012). Thus, future studies could test whether applying tDCS to other regions would produce similar, or even larger, effects on the MI. One potentially unexpected finding is that we did not observe the opposing behavioral effects of anodal and cathodal tDCS that some previous studies have reported. That is, a reduction in slopes was not observed in the right cathode-left anode configuration. Two recent lines of research, however, may explain this finding. First, a recent meta-analysis (Jacobson et al., 2012) indicates that although cathodal stimulation has relatively consistent inhibitory effects on simple motor tasks, the inhibitory effects of cathodal stimulation are much less consistent in behaviorally-complex cognitive tasks. The authors’ explanation for this result is that there is greater capacity for neural compensation in tasks that require the integration of a broad network of brain regions. Unlike the simple motor responses typical of many tDCS motor studies, the complexity of our reaching task to an imagined target may have required a sufficiently large network of brain regions that the inhibitory effects of tDCS may have been compensated for with other nodes in the network. A second possible explanation of why we did not see inhibitory behavioral effects with cathodal stimulation over the right hemisphere is that these effects can be intensity-dependent in motor tasks. For example, although 1 mA cathodal stimulation is typically inhibitory in the motor system, 2 mA cathodal stimulation can result in facilitatory effects similar to 2 mA of anodal stimulation (Batsikadze et al., 2013). Although the effects of 1.5 mA cathodal stimulation on the motor system have not been investigated, it is possible that this current is near the intensity at which the inhibitory and facilitatory effects cross-over. The decision to use 1.5 mA stimulation was made before publication of the work of Batsikadze and colleagues, and we had assumed, naively, that the nature of the tDCS effects would not be intensity-dependent. Nonetheless, we will note that given our goal of applying the results of the present study to future rehabilitation studies, the fact that we reported increased MI effects is more clinically-important than showing that the MI could be reduced. In summary, by establishing the potential benefits of combining MT and tDCS, the results of this study provide promising evidence for a novel, easy to administer, treatment for post-stroke hemiparesis.
Acknowledgements Funding for this study was provided by the William N. Kelley Chair in Neurology to HBC. The authors would also like to thank Adam Woods for his assistance with tDCS training.
51
References Batsikadze, G., Moliadze, V., Paulus, W., Kuo, M.-F., Nitsche, M.A., 2013. Partially non-linear stimulation intensity-dependent effects of direct current stimulation on motor cortex excitability in humans. J. Physiol. 591, 1987–2000. http: //dx.doi.org/10.1113/jphysiol.2012.249730. Brunoni, A.R., Nitsche, M.A., Bolognini, N., Bikson, M., Wagner, T., Merabet, L., Fregni, F., 2012. Clinical research with transcranial direct current stimulation (tDCS): challenges and future directions. Brain Stimul. 5, 175–195. http://dx.doi. org/10.1016/j.brs.2011.03.002. Desmurget, M., Pélisson, D., Rossetti, Y., Prablanc, C., 1998. From eye to hand: planning goal-directed movements. Neurosci. Biobehav. Rev. 22, 761–788. http: //dx.doi.org/10.1016/S0149-7634(98)00004–9. Dohle, C., Pullen, J., Nakaten, A., Kust, J., Rietz, C., Karbe, H., 2009. Mirror therapy promotes recovery from severe hemiparesis: a randomized controlled trial. Neurorehabil. Neural Repair 23, 209–217. http://dx.doi.org/10.1177/ 1545968308324786. Garry, M.I., Loftus, A., Summers, J.J., 2005. Mirror, mirror on the wall: viewing a mirror reflection of unilateral hand movements facilitates ipsilateral M1 excitability. Exp. Brain Res. 163, 118–122. http://dx.doi.org/10.1007/s00221-0052226-9. Hamzei, F., Läppchen, C.H., Glauche, V., Mader, I., Rijntjes, M., Weiller, C., 2012. Functional plasticity induced by mirror training: the mirror as the element connecting both hands to one hemisphere. Neurorehabil. Neural Repair 26, 484–496. http://dx.doi.org/10.1177/1545968311427917. Holmes, N.P., Spence, C., 2005. Visual bias of unseen hand position with a mirror: spatial and temporal factors. Exp. Brain Res. 166, 489–497. http://dx.doi.org/ 10.1007/s00221-005-2389-4. Jacobson, L., Koslowsky, M., Lavidor, M., 2012. tDCS polarity effects in motor and cognitive domains: a meta-analytical review. Exp. Brain Res. 216, 1–10. http: //dx.doi.org/10.1007/s00221-011-2891-9. Läppchen, C.H., Ringer, T., Blessin, J., Seidel, G., Grieshammer, S., Lange, R., Hamzei, F., 2012. Optical illusion alters M1 excitability after mirror therapy: a TMS study. J. Neurophysiol. 108, 2857–2861. http://dx.doi.org/10.1152/ jn.00321.2012. Mahmoudi, H., Borhani Haghighi, A., Petramfar, P., Jahanshahi, S., Salehi, Z., Fregni, F., 2011. Transcranial direct current stimulation: electrode montage in stroke. Disabil. Rehabil. 33, 1383–1388. http://dx.doi.org/10.3109/ 09638288.2010.532283. Michielsen, M.E., Selles, R.W., van der Geest, J.N., Eckhardt, M., Yavuzer, G., Stam, H. J., Bussmann, J.B., 2011. Motor recovery and cortical reorganization after mirror therapy in chronic stroke patients: a phase II randomized controlled trial. Neurorehabil. Neural Repair 25, 223–233. http://dx.doi.org/10.1177/ 1545968310385127. Nitsche, M.A., Paulus, W., 2000. Excitability changes induced in the human motor cortex by weak transcranial direct current stimulation. J. Physiol. 52, 633–639. http://dx.doi.org/10.1111/j.1469-7793.2000.t01-1-00633. Nojima, I., Mima, T., Koganemaru, S., Thabit, M.N., Fukuyama, H., Kawamata, T., 2012. Human motor plasticity induced by mirror visual feedback. J. Neurosci. 32, 1293–1300. http://dx.doi.org/10.1523/JNEUROSCI.5364-11.2012. Ramachandran, V.S., Altschuler, E.L., 2009. The use of visual feedback, in particular mirror visual feedback, in restoring brain function. Brain 132, 1693–1710. http: //dx.doi.org/10.1093/brain/awp135. Saleh, S., Adamovich, S.V., Tunik, E., 2013. Mirrored feedback in chronic stroke: recruitment and effective connectivity of ipsilesional sensorimotor networks. Neurorehabil. Neural Repair 28, 344–354. http://dx.doi.org/10.1177/ 1545968313513074. Selles, R.W., Michielsen, M.E., Bussmann, J.B.J., Stam, H.J., Hurkmans, H.L., Heijnen, I., Ribbers, G.M., 2014. Effects of a mirror-induced visual illusion on a reaching task in stroke patients: implications for mirror therapy training. Neurorehabil. Neural Repair 28, 652–659. http://dx.doi.org/10.1177/1545968314521005. Thieme, H., Mehrholz, J., Pohl, M., Behrens, J., Dohle, C., 2013. Mirror therapy for improving motor function after stroke. Stroke 44, e1–e2. http://dx.doi.org/ 10.1161/STROKEAHA.112.673087. Van Beers, R.J., Sittig, A.C., Gon, J.J.D. van der, 1999. Integration of proprioceptive and visual position-information: an experimentally supported model. J. Neurophysiol. 81, 1355–1364.
Contents lists available at ScienceDirect
Neuropsychologia journal homepage: www.elsevier.com/locate/neuropsychologia
Enhancing the mirror illusion with transcranial direct current stimulation Steven A. Jax a,n,1, Diana L. Rosa-Leyra a, H. Branch Coslett b a b
Moss Rehabilitation Research Institute, Elkins Park, PA, USA Department of Neurology, University of Pennsylvania School of Medicine, USA
art ic l e i nf o
a b s t r a c t
Article history: Received 25 November 2014 Received in revised form 3 February 2015 Accepted 17 March 2015 Available online 19 March 2015
Visual feedback has a strong impact on upper-extremity movement production. One compelling example of this phenomena is the mirror illusion (MI), which has been used as a treatment for post-stroke movement deficits (mirror therapy). Previous research indicates that the MI increases primary motor cortex excitability, and this change in excitability is strongly correlated with the mirror’s effects on behavioral performance of neurologically-intact controls. Based on evidence that primary motor cortex excitability can also be increased using transcranial direct current stimulation (tDCS), we tested whether bilateral tDCS to the primary motor cortices (anode right-cathode left and anode left-cathode right) would modify the MI. We measured the MI using a previously-developed task in which participants make reaching movements with the unseen arm behind a mirror while viewing the reflection of the other arm. When an offset in the positions of the two limbs relative to the mirror is introduced, reaching errors of the unseen arm are biased by the reflected arm’s position. We found that active tDCS in the anode right-cathode left montage increased the magnitude of the MI relative to sham tDCS and anode left-cathode right tDCS. We take these data as a promising indication that tDCS could improve the effect of mirror therapy in patients with hemiparesis. & 2015 Elsevier Ltd. All rights reserved.
Keywords: Mirror therapy Mirror feedback Reaching Transcranial direct current stimulation
1. Introduction The influence of visual feedback on upper extremity movement production is well documented (Desmurget et al., 1998). One compelling application of this finding comes from mirror therapy (MT) as it is used to treat post-stroke hemiparesis. Originally developed as a treatment for phantom limb pain (Ramachandran and Altschuler, 2009), MT involves the patient being seated in front of a vertically-oriented mirror (Fig. 1). The patient places her unimpaired arm on the reflective side of the mirror and her impaired arm behind the mirror so that it is hidden. When the mirror is placed midway between the two limbs, movements of the unimpaired limb (viewed in the mirror) appear in the same location as the impaired limb. Thus, the MT setup creates a compelling illusion in which movements of the impaired arm behind the mirror appear to be made as effectively as the unimpaired arm. Although previous clinical research indicates that MT is efficacious (Thieme et al., 2013) the underlying neural changes associated with MT are poorly understood. Because MT-like effects can also be observed in neurologically-intact controls (which we will n
Corresponding author: E-mail address: [email protected] (S.A. Jax). 1 50 Township Line Rd., Elkins Park, PA, 19027, USA.
http://dx.doi.org/10.1016/j.neuropsychologia.2015.03.017 0028-3932/& 2015 Elsevier Ltd. All rights reserved.
term the “mirror illusion,” or MI, to differentiate it from the therapeutic use of the mirror setup), several studies have examined the possible neural mechanisms of the MI. Many of these studies, as well as studies of stroke survivors, suggest that mirror feedback modifies functioning within the primary motor cortex of the hemisphere that controls the arm behind the mirror (the lesioned hemisphere in stroke survivors). In controls, the MI increases primary motor cortex excitability (Garry, et al., 2005) and this change in excitability is strongly correlated with the mirror’s influence on movements of the arm behind the mirror (Nojima et al., 2012). These changes in excitability may come about via intra- or inter-hemispheric neuroplasticity (Hamzei et al., 2012; Nojima et al., 2012). In stroke survivors, effective connectivity between the primary motor cortex and the somatosensory cortex in the lesioned hemisphere is increased under single-session MT conditions (Saleh et al., 2013) and completing 6 weeks of MT increases BOLD activation of the lesioned hemisphere’s primary motor cortex relative to the intact hemisphere’s primary motor cortex (Michielsen et al., 2011). While many studies indicate that the mirror feedback in the MI and during MT primarily changes neural functioning within the hemisphere controlling the hand behind the mirror, conflicting findings have been reported. For example, increased reliance on regions within the hemisphere controlling the visible hand have
S.A. Jax et al. / Neuropsychologia 71 (2015) 46–51
47
“Predictions” section below. We used this task as an index of the effect that applying bilateral primary motor cortex tDCS had on the MI. Previous research indicates that anodal tDCS increases primary motor cortex excitability whereas cathodal tDCS decreases excitability (Nitsche and Paulus, 2000). If tDCS’ effects on primary motor cortex excitability are similar to those reported from the MI (Garry et al., 2005; Nojima et al., 2012) then applying tDCS during the task should affect the MI. Predictions about the effect of tDCS depend on the presumed site of the MI effect and the tDCS montage. If one assumes that the MI results in increased excitability within the hemisphere primarily controlling the hand behind the mirror (the right hemisphere in the present study), then applying anodal stimulation to the right hemisphere and cathodal stimulation to the left hemisphere should increase the MI. Similarly, reversing the polarity of stimulation (right cathode-left anode) should reduce the MI. In contrast, if one assumes that the MI results in increased excitability within the hemisphere controlling the hand in front of the mirror (the left hemisphere in the present study), then the MI should be increased in the right cathode-left anode montage and decreased in the right anode-left cathode montage. The present study tested these contrasting predictions.
2. Methods 2.1. Participants Fig. 1. Reaching task setup. The participant’s right hand covers the near (R1) start location, while the dot indicating the R2 start location is visible to the right of the hand. The white box above and to the left of the mirror highlights the two dots that indicated the position of the imagined target locations along the platform, which the participant reached for with the unseen left hand.
Twelve college-aged young adults [mean (s.d.) age 21.3 (2.4) years old; 5 females] completed the study. All participants were recruited through the University of Pennsylvania and received monetary compensation for their participation. 2.2. tDCS stimulation
been reported using TMS (Läppchen et al., 2012) and neuroimaging (Hamzei et al., 2012) methodologies. In addition, all of the studies cited in the previous paragraph except one (Michielsen et al., 2011) involved no movement of the hand behind the mirror. These conditions differ from those typically used during MT, in which patients attempt to move both limbs simultaneously (Dohle et al., 2009; Michielsen et al., 2011). It is unknown whether the neural changes associated with uni- and bi-manual MT/MI are different, but one study in patients suggests that these conditions can lead to different behavioral outcomes (Selles et al., 2014). Thus, to date there is conflicting evidence about the neural mechanisms of MT. Resolving this conflict may have significant therapeutic significance, as understanding these mechanisms could be used to augment MT’s endogenous neural changes with exogenous non-invasive brain stimulation such as transcranial direct current stimulation (tDCS). The present study was designed to provide an initial step towards this goal by testing whether applying tDCS to the two primary motor cortices would affect the MI in neurologically-intact controls. Given our goal of applying the results of this study to post-stroke rehabilitation, we wanted to examine the effects of tDCS on motor performance when bimanual movements were sometimes required. To do so, we utilized a task developed by Holmes and Spence (2005) to elicit the MI. Participants began by making symmetric bilateral movements within the mirror setup. Then, participants made a reaching movement with the unseen left arm (behind the mirror) while viewing the reflection of the stationary right arm. When an offset in the positions of the two limbs relative to the mirror was introduced, the unseen hand’s trajectory was biased as if it were starting nearer the position of the mirror-reflected right hand, an effect not observed when the mirror was covered. A more detailed description of how the task allowed quantification of the MI will be provided in the
Each participant completed three separate single-day testing sessions, each of which employed a different tDCS montage. In all sessions, two 5 7 cm2 saline soaked electrodes were positioned over C3 and C4 of the International 10–20 EEG placement system. Within each session, a Magstim tDCS stimulator provided 1.5 mA of current in one of three montages: (1) right cathode-left anode, (2) left cathode-right anode or (3) sham. In the sham montage the same electrode locations were used (half of participants had right cathode-left anode, half had left cathode-right anode) but with a 30 s ramp up then 30 s ramp down of 1.5 mA simulation to provide tactile sensations similar to the other montages. Stimulation began at the start of each reaching task and continued until the task was complete (approximately 20 min). Each session was separated by at least two weeks to minimize any carry-over effects of both tDCS and the reaching task. 2.3. Reaching task Within each session, participants completed the same reaching task. Participants sat slightly to the right of the vertically-oriented mirror (30 cm 30 cm) so they could see the reflection of their right arm in the mirror (Fig. 1). Vision of the left arm was occluded by the mirror as well as by black fabric draped between the front of the apparatus and the participant’s left shoulder. Each trial began when the experimenter instructed the participant to place his or her right index finger on one of two starting positions indicated by a red and blue dot on the platform (9 and 16 cm to the right of the mirror; approximately 19 cm in front of the participant’s body). Next, the experimenter moved the participant’s left arm to one of the four left hand start positions (9, 16, 23 and 30 cm to the left of the mirror; approximately 19 cm in front of the
48
S.A. Jax et al. / Neuropsychologia 71 (2015) 46–51
participants body). Once in position, participants were instructed to synchronously flex and extend both wrists in time with a 120 bpm metronome for 8 s while gazing at the right arm in the mirror. These wrist movements were included to elicit the mirror illusion in a consistent fashion before each trial, a technique similar to the one used by Holmes and Spence (2005). Next, participants used their left arms to reach forward and touch an imagined target located on the platform directly below either a red or blue dot visible on the frame of the apparatus (highlighted with a white box in Fig. 1). The target to which subjects reached was the same color as the right hand start dot. While participants performed the task, movements of the left index finger were recorded using an Ascension TrakSTAR kinematic recording system sampling at 150 Hz so that endpoint reaching errors could be measured. A sensor was also placed on the right index finger so that tactile sensation was similar across the hands. Each block included 16 trials randomly comprising two repetitions of all combinations of the two right hand starting positions (R1, R2) and four left hand starting positions (L1, L2, L3, L4). Each session included four blocks, two with the mirror visible and two with the mirror completely covered so that the reflection of the right arm was not visible. The order of mirror and covered blocks was counterbalanced within the session using an ABBA order. The status of the session's first block (mirror or covered) was counterbalanced across participants and sessions. Thus, within each session participants completed 64 trials, or four repetitions of the 16 possible combinations of the two mirror conditions (mirror, covered), the two right hand starting positions, and the four left hand starting positions. 2.4. Predictions Fig. 2 illustrates how the experimental method allowed us to
determine the effect that the mirror-reflected right hand had on reaching in the unseen left hand (the MI). Predictions are best illustrated by considering the left hand’s endpoint errors along the left-right axis under two extreme possible outcomes: when the mirror-reflected right hand has no effect on the left hand (left column of Fig. 2), and when the mirror-reflected right hand defines the location of the left hand in left hemispace (middle column). In the first example (A and B), the left hand (left triangle at position L1 in A) starts the trial closer to the mirror than the right hand (right triangle at position R2; note that targets R1 and L1 are equidistant from the mirror, as are R2 and L2). If the position of mirror-reflected right hand has no effect on the left hand’s error, the left hand should move directly to the target (solid arrow in A). In contrast, if the reflected right hand’s position defines the position of the left hand in left hemispace, an accurately planned movement would result in the left hand moving parallel to the mirror directly to the target (dashed arrow in B). However, since the left hand is actually located closer to the mirror than the right hand, executing the same movement parallel to the mirror from the actual starting position would result in an endpoint error that is to the right of the target (solid arrow in B). Thus, if the reflected right hand’s position is taken to be the position of the left hand in left hemispace, endpoint errors will be biased towards the starting position of the left hand. A second example (C and D) illustrates this directional effect. In contrast to the previous example, the left hand (at L3) starts further from the mirror than the right hand (at R2). As in the previous example, if the mirror has no effect the participant will reach directly to the target. However, if the reflected right hand is taken to indicate the position of the left hand in left hemispace, the left hand should again move parallel to the mirror directly to the target (dashed arrow in D). Because the left hand is further from the mirror than the right hand, the endpoint error is biased to the
Fig. 2. Examples of predicted results under the hypothesis that the mirror will not affect left-hand reaching (left column; A, C, E) and under the hypothesis that the reflected right hand’s position will be taken to be the position of the left hand in left hemispace (middle column; B, D, F). In Panels B, D, and F, dashed arrows indicate the proper movement vector if the left hand position was assumed to be in the location of the mirror reflected right hand, and solid arrows indicate the corresponding movement path from the actual starting position if that vector were to be executed. Predictions across all left hand start locations are summarized in Panel G for the R2 start position. See Predictions section in the text for details.
S.A. Jax et al. / Neuropsychologia 71 (2015) 46–51
49
left of the target (solid arrow in D). Like the first example, the leftright endpoint errors are biased towards the starting position of the left hand. The magnitude of this bias is further increased when the offset between the hands is increased (E and F). To summarize the predictions of these two extreme possible outcomes (no effect of mirror and total effect of mirror), one can plot the expected left-right endpoint errors as a function of the four left hand start positions for a single right-hand start position (e.g. R2, as illustrated in this figure). Under the outcome in which the mirror will have no effect, reaching errors will be observed (because reaching is done without visual feedback), but the left-right bias of the errors is predicted to be near zero and be unaffected by changes in the left hand start position. That is, a line connecting the error bias of the four left hand start positions should have a slope near zero (dashed line in G). In contrast, under the outcome that the mirror-reflected position of the right hand will be taken as the position of the left hand in left hemispace (total effect of the mirror), endpoint errors will be biased to the right when the left hand starts closer to the mirror than the right hand (B) and biased to the left when the left hand starts further from the mirror than the right hand (D and F). Therefore, a positive slope in the line connecting the errors of the four left hand start positions would indicate evidence for the MI (solid line in G), with the magnitude of the slope quantifying the magnitude of the mirror’s effect. Based on the results of Holmes and Spence (2005) we predicted that the slopes of the error lines would be greater in the mirror conditions than the covered conditions. In addition, we tested whether slopes in the mirror condition would be affected by tDCS. As described in the Introduction, if the MI results in increased excitability within the hemisphere controlling the hand behind the mirror (the right hemisphere in the present study), then slopes in the mirror condition should increase in the anode right condition and decrease in the anode left condition. In contrast, if one assumes that the MI results in increased excitability within the hemisphere controlling the hand in front of the mirror (the left hemisphere in the present study), then the opposite pattern should be observed. That is, slopes should increase in the anode left montage and decrease in the anode right montage.
3. Results The average left-right endpoint error data for all conditions are shown in Fig. 3. In this figure, two central results can be observed. First, slopes in the error lines for all conditions were steeper in the mirror conditions (solid lines) than the covered conditions (dashed lines). Second, mirror condition slopes of the right-anode condition were steeper than the other two tDCS conditions. To confirm these results using inferential statistics in the most direct and transparent way2, we used regression analysis to compute the best-fitting regression line predicting the left-right endpoint errors (in cm) for the L1, L2, L3, and L4 starting positions with the corresponding distances the left hand was from the 2
An alternative analysis approach that we considered but rejected was analyzing the endpoint error data shown in Fig. 3 with a 4-way ANOVA with the factors tDCS (anode right, anode left, sham), Right hand start location (1,2), Left hand start location (1, 2, 3, 4), and Mirror (mirror, covered). Using this approach, one would need to examine interactions to test the central hypotheses about slopes. For example, one would look for an interaction between the Left hand start location factor and the Mirror factor to show that slopes were steeper in the mirror condition than the covered condition. The limitation of this ANOVA approach is that testing for an interaction is, graphically, only a test to determine if the lines are not completely parallel to one another. Thus, although all differences in slope would lead to an interaction, not all interactions would indicate differences in slope. Therefore, we choose to directly compute slope values and then analyze those slopes using an ANOVA.
Fig. 3. Mean left-right endpoint error data (7 1 s.e.) for all conditions. Separate panels illustrate results from the three tDCS conditions (left anode, right anode, and sham). Within panels, errors are shown separately for the mirror (solid line) and covered (dashed line) conditions. Lines connect errors for the four possible left hand start locations (L1, L2, L3, and L4) separately for the two right hand start locations (R1 or R2).
mirror at the start of the trial (9, 16, 23 and 30 cm, respectively). Note that the value of the slope indicates the proportion of the possible MI effect that was observed because it measures how many centimeters one would expect endpoint error to change with an additional 1 cm shift in left hand start location. Thus, a
50
S.A. Jax et al. / Neuropsychologia 71 (2015) 46–51
Fig. 4. Mean slope values ( 71 s.e.) for the mirror and covered conditions, reported separately for the three tDCS conditions (separate bars) and two right hand start locations (R1 and R2).
slope of 0 indicates 0% of the possible MI effect observed, and a slope of 1 indicates 100% of the possible MI effect. Regressions were completed at the participant level, and for each participant 12 best-fitting lines were computed for each combination of tDCS (anode right, anode left, sham), mirror (mirror, covered), and right hand start position (1, 2). The slopes of these 144 lines (12 participants 12 condition combinations per participant) were then analyzed using a 3 (tDCS: anode right, anode left, (Right hand start: 1,2) 2 (Mirror: mirror, covered) ANOVA, with all factors manipulated within participants. The two sham configurations were grouped because analyses indicated no differences between the montages. The results of the slope analysis ANOVA are shown in Fig. 4. There was a main effect of Mirror, F(1,11) ¼43.72, p o.001, with average slopes being higher in the mirror conditions than the covered conditions (slope of 0.099 vs. 0.033, respectively). There was also a main effect of tDCS, F(2,22) ¼4.68, p¼ .020. Bonferronicorrected follow-up t-tests indicated that average slopes were significantly higher (p¼ .025) in the anode right condition than the sham condition (slope of 0.080 vs. 0.059), and that average slopes in the anode right condition were marginally (p ¼.067) higher in the anode right than the anode left condition (slope of 0.080 vs. 0.058). Most importantly, the Mirror and tDCS factors interacted significantly, F(2,22) ¼ 4.25, p ¼ .027. When the effects of tDCS were analyzed separately for the mirror and covered conditions, there was no effect of tDCS in the covered condition, F(2,22) ¼.056, p ¼.946, whereas this effect was significant in the mirror condition, F(2,22) ¼8.52, p¼ .002. Bonferroni-corrected follow-up t-tests indicated that average slopes in the anode right condition were significantly higher than the anode left condition (slope of 0.126 vs. 0.087, p ¼.037) and the sham condition (slope of 0.126 vs. 0.083, p ¼.008). No effects of Right hand start, nor any interactions between this factor and other factors, were observed (p's 4.37).
4. Discussion In this study we showed for the first time that the mirror illusion could be increased using noninvasive brain stimulation. When concurrent anodal tDCS was delivered to the primary motor cortex controlling the unseen arm behind the mirror (along with cathodal stimulation to the primary motor cortex controlling the visible arm), reaching errors of the unseen arm were more biased by the reflected arm’s position than when sham stimulation or stimulation with the opposite polarity was delivered. This effect of tDCS was not observed in the control condition in which the
mirror was covered. Thus, tDCS specifically affected the way the mirrored visual feedback was utilized by the motor system, and the effects of tDCS we observed cannot be ascribed to general changes in reaching with the left arm. Critical for our goal of applying this result to post-stroke motor rehabilitation, the tDCS montage that increased the MI was the same montage that has been used therapeutically to treat hemiparesis under non-mirrortherapy conditions (Mahmoudi et al., 2011). Thus, this study provides a promising possibility for improved post-stroke rehabilitation by combining MT and tDCS in future studies. In the remainder of the Discussion section, we will discuss (1) the role that proprioception may have played in our results and (2) the implications of our finding for the neural substrates of the mirror illusion and mirror therapy. In the mirror illusion, the visually-derived estimate of the location of the arm behind the mirror (the left arm in the present study) is manipulated. It is important to note that proprioception also plays an important role in limb position estimation (Van Beers et al. 1999). The influence of proprioception to the MI is clearly observed in two results in the present study. First, the magnitude of the MI in the mirror condition was a fraction of the total possible MI, defined as the predicted results if the mirror-reflected right hand was taken to be the location of the left hand in left hemispace (see Fig. 2). Across the tDCS conditions, the slopes of the error bias were between .08 and .14, or 8–14% of the possible MI. These slope were lower than slopes in comparable conditions of Holmes and Spence (2005), which were approximately .30 or 30% of the possible MI. Further replication will be needed to better understand this difference in MI magnitude from two studies using highly similar, but not identical, methods. Nonetheless, the effects of the mirror were consistent, but did not completely “trick” the participant into believing the reflected arm truly indicated the left arm’s location. This incomplete effect is likely due to the proprioceptive signal indicating the left arm’s actual location, which can then be modulated by the visual illusion provided by the mirror. The limited effect of the mirror may have also been due to the relatively large possible asymmetries in the positions of the two arms. Under the most extreme condition, the right hand started 9 cm from the mirror while the left hand started 30 cm from the mirror. The MI would have to quite strong to completely overcome the proprioceptive signals about the hidden arm's position with such an offset. The effects of proprioception were also likely observable in the slightly positive slopes in the covered conditions, which indicated that the right hand’s position affected reaching with the left hand, even when the mirrored visual feedback was not present. This finding may have been caused by the motor system’s tendency to assume proprioceptive symmetry between the limbs, such that the proprioceptively-sensed location of the right arm would provide some influence of the estimate of the left arm’s position. Thus, at least some of the MI may be caused by proprioceptive mirroring. However, the visual input clearly matters given that slopes were at least twice as great in the mirror condition than the covered condition. If proprioceptive symmetry were the sole cause of the MI, results would be predicted to be similar across the mirror and covered conditions. Given the role that proprioception can play in this task, it is logical to ask whether applying tDCS centered on the primary motor cortex may have also stimulated the primary somatosensory cortex, and therefore also affected proprioceptive processing. Given the non-focality of tDCS, this is a possibility. However, the effects of tDCS were only seen in the mirror condition. Thus, it is unlikely that our results were due to changes in the way that all proprioceptive information is processed because such changes would have affected both the mirror and covered conditions similarly. However, it is possible that tDCS affected how visual and
S.A. Jax et al. / Neuropsychologia 71 (2015) 46–51
proprioceptive information were integrated. Future studies with more focal forms of stimulation (such as transcranial magnetic stimulation) could investigate this issue further. Relatedly, even though we applied tDCS to the primary motor cortices, the neural changes responsible for the modulation of the MI need not necessarily emanate solely from the primary motor cortex. Visuallyguided upper extremity movements are controlled by a network of brain regions, and the effects of tDCS are known to extend beyond the regions underneath the electrodes (Brunoni et al., 2012). Thus, future studies could test whether applying tDCS to other regions would produce similar, or even larger, effects on the MI. One potentially unexpected finding is that we did not observe the opposing behavioral effects of anodal and cathodal tDCS that some previous studies have reported. That is, a reduction in slopes was not observed in the right cathode-left anode configuration. Two recent lines of research, however, may explain this finding. First, a recent meta-analysis (Jacobson et al., 2012) indicates that although cathodal stimulation has relatively consistent inhibitory effects on simple motor tasks, the inhibitory effects of cathodal stimulation are much less consistent in behaviorally-complex cognitive tasks. The authors’ explanation for this result is that there is greater capacity for neural compensation in tasks that require the integration of a broad network of brain regions. Unlike the simple motor responses typical of many tDCS motor studies, the complexity of our reaching task to an imagined target may have required a sufficiently large network of brain regions that the inhibitory effects of tDCS may have been compensated for with other nodes in the network. A second possible explanation of why we did not see inhibitory behavioral effects with cathodal stimulation over the right hemisphere is that these effects can be intensity-dependent in motor tasks. For example, although 1 mA cathodal stimulation is typically inhibitory in the motor system, 2 mA cathodal stimulation can result in facilitatory effects similar to 2 mA of anodal stimulation (Batsikadze et al., 2013). Although the effects of 1.5 mA cathodal stimulation on the motor system have not been investigated, it is possible that this current is near the intensity at which the inhibitory and facilitatory effects cross-over. The decision to use 1.5 mA stimulation was made before publication of the work of Batsikadze and colleagues, and we had assumed, naively, that the nature of the tDCS effects would not be intensity-dependent. Nonetheless, we will note that given our goal of applying the results of the present study to future rehabilitation studies, the fact that we reported increased MI effects is more clinically-important than showing that the MI could be reduced. In summary, by establishing the potential benefits of combining MT and tDCS, the results of this study provide promising evidence for a novel, easy to administer, treatment for post-stroke hemiparesis.
Acknowledgements Funding for this study was provided by the William N. Kelley Chair in Neurology to HBC. The authors would also like to thank Adam Woods for his assistance with tDCS training.
51
References Batsikadze, G., Moliadze, V., Paulus, W., Kuo, M.-F., Nitsche, M.A., 2013. Partially non-linear stimulation intensity-dependent effects of direct current stimulation on motor cortex excitability in humans. J. Physiol. 591, 1987–2000. http: //dx.doi.org/10.1113/jphysiol.2012.249730. Brunoni, A.R., Nitsche, M.A., Bolognini, N., Bikson, M., Wagner, T., Merabet, L., Fregni, F., 2012. Clinical research with transcranial direct current stimulation (tDCS): challenges and future directions. Brain Stimul. 5, 175–195. http://dx.doi. org/10.1016/j.brs.2011.03.002. Desmurget, M., Pélisson, D., Rossetti, Y., Prablanc, C., 1998. From eye to hand: planning goal-directed movements. Neurosci. Biobehav. Rev. 22, 761–788. http: //dx.doi.org/10.1016/S0149-7634(98)00004–9. Dohle, C., Pullen, J., Nakaten, A., Kust, J., Rietz, C., Karbe, H., 2009. Mirror therapy promotes recovery from severe hemiparesis: a randomized controlled trial. Neurorehabil. Neural Repair 23, 209–217. http://dx.doi.org/10.1177/ 1545968308324786. Garry, M.I., Loftus, A., Summers, J.J., 2005. Mirror, mirror on the wall: viewing a mirror reflection of unilateral hand movements facilitates ipsilateral M1 excitability. Exp. Brain Res. 163, 118–122. http://dx.doi.org/10.1007/s00221-0052226-9. Hamzei, F., Läppchen, C.H., Glauche, V., Mader, I., Rijntjes, M., Weiller, C., 2012. Functional plasticity induced by mirror training: the mirror as the element connecting both hands to one hemisphere. Neurorehabil. Neural Repair 26, 484–496. http://dx.doi.org/10.1177/1545968311427917. Holmes, N.P., Spence, C., 2005. Visual bias of unseen hand position with a mirror: spatial and temporal factors. Exp. Brain Res. 166, 489–497. http://dx.doi.org/ 10.1007/s00221-005-2389-4. Jacobson, L., Koslowsky, M., Lavidor, M., 2012. tDCS polarity effects in motor and cognitive domains: a meta-analytical review. Exp. Brain Res. 216, 1–10. http: //dx.doi.org/10.1007/s00221-011-2891-9. Läppchen, C.H., Ringer, T., Blessin, J., Seidel, G., Grieshammer, S., Lange, R., Hamzei, F., 2012. Optical illusion alters M1 excitability after mirror therapy: a TMS study. J. Neurophysiol. 108, 2857–2861. http://dx.doi.org/10.1152/ jn.00321.2012. Mahmoudi, H., Borhani Haghighi, A., Petramfar, P., Jahanshahi, S., Salehi, Z., Fregni, F., 2011. Transcranial direct current stimulation: electrode montage in stroke. Disabil. Rehabil. 33, 1383–1388. http://dx.doi.org/10.3109/ 09638288.2010.532283. Michielsen, M.E., Selles, R.W., van der Geest, J.N., Eckhardt, M., Yavuzer, G., Stam, H. J., Bussmann, J.B., 2011. Motor recovery and cortical reorganization after mirror therapy in chronic stroke patients: a phase II randomized controlled trial. Neurorehabil. Neural Repair 25, 223–233. http://dx.doi.org/10.1177/ 1545968310385127. Nitsche, M.A., Paulus, W., 2000. Excitability changes induced in the human motor cortex by weak transcranial direct current stimulation. J. Physiol. 52, 633–639. http://dx.doi.org/10.1111/j.1469-7793.2000.t01-1-00633. Nojima, I., Mima, T., Koganemaru, S., Thabit, M.N., Fukuyama, H., Kawamata, T., 2012. Human motor plasticity induced by mirror visual feedback. J. Neurosci. 32, 1293–1300. http://dx.doi.org/10.1523/JNEUROSCI.5364-11.2012. Ramachandran, V.S., Altschuler, E.L., 2009. The use of visual feedback, in particular mirror visual feedback, in restoring brain function. Brain 132, 1693–1710. http: //dx.doi.org/10.1093/brain/awp135. Saleh, S., Adamovich, S.V., Tunik, E., 2013. Mirrored feedback in chronic stroke: recruitment and effective connectivity of ipsilesional sensorimotor networks. Neurorehabil. Neural Repair 28, 344–354. http://dx.doi.org/10.1177/ 1545968313513074. Selles, R.W., Michielsen, M.E., Bussmann, J.B.J., Stam, H.J., Hurkmans, H.L., Heijnen, I., Ribbers, G.M., 2014. Effects of a mirror-induced visual illusion on a reaching task in stroke patients: implications for mirror therapy training. Neurorehabil. Neural Repair 28, 652–659. http://dx.doi.org/10.1177/1545968314521005. Thieme, H., Mehrholz, J., Pohl, M., Behrens, J., Dohle, C., 2013. Mirror therapy for improving motor function after stroke. Stroke 44, e1–e2. http://dx.doi.org/ 10.1161/STROKEAHA.112.673087. Van Beers, R.J., Sittig, A.C., Gon, J.J.D. van der, 1999. Integration of proprioceptive and visual position-information: an experimentally supported model. J. Neurophysiol. 81, 1355–1364.
