To our knowledge, this study is the first to compare the cerebral activation induced by conventional mirror therapy (MT) with that induced by video therapy (VT). In VT, the visualised movements are pre-recorded and projected onto a large screen positioned in front of the individual. Compared with conventional MT, VT could provide a higher quality illusion and encourage attention to visual feedback. This new modality could therefore improve the effectiveness of MT by optimising the mechanisms that induce neuroplasticity. In this study we focused on the regions of interest that are involved in MT, the precuneus (PC), primary motor cortex (M1) and premotor cortex (PMC). We explored mirror tasks with no movement intention, which allowed us to specifically assess the effect of visual feedback. Our main aim was to evaluate the difference in activation of the PC between the two techniques since PC activation could be correlated with the effectiveness of the technique [19]. As we had hypothesized, we found greater activation of the PC contralateral to visual feedback during VT than during conventional MT. We also showed that activation of M1 and PMC contralateral to visual feedback was more extensive in VT than in MT.

The involvement of the PC in MT has been previously widely demonstrated [16, 17, 19, 34]. The PC plays a major role in visual processing and self-perception [20], and more particularly in the perception of the hand [35]. We initially hypothesized that during the MT and the VT tasks, the PC would be activated if the participants perceived the illusion of seeing their own hand during the visual feedback situation, and we hypothesised that the quality of the illusion would be greater with VT. To verify our hypothesis, we assessed the participants’ perceptions of the quality of the illusion (impression that the hand was moving, tingling, sensations of contraction, etc.), as suggested by Rossiter [36]. We found that the perception of the illusion did not differ between the VT and MT tasks and that it was not correlated with the level of PC activation. Illusion perception is subjective and difficult to assess, in particular because no validated scales exist for that purpose. However, some studies have found PC activation during VT without a real illusion [16, 19]. In those studies, the screen was located at a distance from the participant that did not allow visual continuity between the upper limb and the visual feedback. Thus, we cannot conclude that the greater PC activation found in this study with VT than MT was the result of a higher quality movement illusion with VT.

Another explanation for these results could relate to the role of the PC in attentional tasks. When a person is focused on a given task, the PC may be recruited to support the cognitive engagement. A functional MRI (fMRI) study on 10 healthy individuals showed that the PC was activated during a visuospatial attention task [37]. In addition, a meta-analysis showed that PC lesions could cause spatial hemineglect [38]. The PC therefore seems to be particularly involved in visual attention processes. In our study, the lack of an associated motor task during the VT task may have focused attention on the visual feedback to a greater extent than the MT task. Thus, the more extensive PC activation during VT than MT may have been due to a higher level of attention.

Moreover, in our study PC activation was not only contralateral to the side of feedback but bilateral. This could be explained by the fact that the function of the PC is not as lateralized as that of other brain structures. Indeed, although several MT studies have found that PC activation was strictly contralateral in response to visual feedback [16, 18, 19], two motor imaging studies reported that lateralization of the PC was random across individuals. This result is interesting insofar as in the event of a hemispheric lesion including the PC, its activity could be compensated for by the contralesional PC.

The activity of the PC seems important for rehabilitation, as it could have a predominant role in the stimulation of neuroplasticity. Indeed, it has been shown that the PC is closely connected to the motor cortex [39]. The motor cortex is often damaged after a stroke in individuals with residual upper limb impairment. Activating the PC during rehabilitation could therefore stimulate the ipsilesional M1. Based on these results, it is possible that VT, by improving recruitment of precunei, is an effective technique for improving neuroplasticity and therefore motor recovery in stroke patients. These hypotheses will need to be verified in future clinical studies.

First, we only found activation of the left M1 during the MT and control tasks, i.e., the tasks that required motor activity of the right hand. These results are in line with the classical literature regarding the contralateral cortical control of motor activity [40].

Second, we found activation of the right M1 for the tasks involving mirror feedback (MT and VT). These results are also in agreement with the existing literature [12, 13]. However, although we didn’t find any statistical difference in the task comparison for this region, our results show a more extensive activation during VT task than during MT task (for this ROI, one channel was activated during MT task and two channels during VT task). We previously argued that the lack of a motor task during the VT is interesting because it encourages attention on the visual feedback. This absence of a motor task could also explain the difference in M1 activation between VT and MT for two reasons. First, the VT used here, due to the absence of a motor task, could be considered as action observation (AO) therapy associated with visual illusion. A study has shown that AO regulates interhemispheric interaction, with a facilitatory effect on the M1 contralateral to the observed movement and an inhibitory effect on the M1 ipsilateral [41]. However, the VT used here also provided the visual illusion of movement of the own limb. One study showed that AO was able to induce neuroplasticity on M1 only if associated with an illusion of movement (kinaesthetic in the study) [42]. Therefore, the VT used here could have a facilitatory effect on the ipsilesional side (in the case of a stroke) and stimulate neuroplasticity in the M1. Second, the less extensive activation during MT may result from interhemispheric inhibition induced by the right-hand motor task. Indeed, it has been shown that unilateral movement leads to inhibition of the ipsilateral hemisphere via the transcallosal pathway [43]. Interhemispheric balance is altered by stroke [44], even in a resting situation [45], and MT in particular helps to restore this balance [46]. Therefore, it is likely that the difference found in this study would be more marked in a group of people after stroke. However, these results need to be interpreted with caution. Indeed, even though activation involves more channels in VT and is consequently more extensive, our results show no statistically significant difference between the two techniques for the whole ROI. It has also been shown that these interhemispheric interactions can be both inhibitory and facilitatory, depending on stimulus intensity [47, 48]. Here, our results seem to show that the stimulus (the motor task) had an inhibitory effect, since activation was more extensive in the absence of motor task. However, it might be interesting to investigate a possible facilitating effect of the motor task during MT on activation of the M1 contralateral to visual feedback by varying the intensity of the task.

To resume our results concerning M1, they can be explained by a facilitatory effect of VT or an inhibitory effect of MT. In both cases, the absence of a motor task in VT could lead to better stimulation of M1.

Finally, our results showed activation of the PMC contralateral to the visual feedback only during the VT task. This activation was limited to a single channel. This could correspond to activation of mirror neurons located in the ventral part of the PMC [49]. This system is particularly involved in action observation therapy [50]. As mentioned above, the VT used here could be considered as 1st-person AO, which leads to greater activation of mirror neurons than 3rd-person AO [51]. The activation found here could therefore be linked solely to movement observation (with no associated motor task) and would not be dependent on illusion perception, which is the basis of mirror therapy.

In summary, contralateral to the visual feedback, our results show a greater activation of PC during VT compared to MT and an activation of the motor cortex during MT and VT, but this activation was more extensive during VT. These results are mainly explained by the absence of a motor task during VT, which favours increased attention to the visual feedback (greater activation of PC) and possibly reduces interhemispheric inhibition mechanisms (larger activation of M1). Therefore, VT appears to optimise recruitment of the parietofrontal motor network compared with MT [52]. This network induces neuroplasticity after a brain lesion. Indeed, this network is more activated during a motor task in people after stroke than in healthy subjects [53]. Our results therefore suggest that VT may be clinically more effective than MT because of a greater stimulation of neuroplasticity, but this needs to be demonstrated in clinical studies of people after stroke. While the literature shows similar activation patterns between healthy subjects and stroke patients in the exploration of MT mechanisms [13, 14, 16, 17, 34], these mechanisms may be impacted by the location of the lesion. A study of 36 stroke patients showed that the clinical efficacy of MT was linked to the integrity of dorsal and ventral streams [54]. In view of the results of Brunetti et al. [19]. , , we can also assume that a lesion of the PC would also impact the clinical efficacy of MT. Thus, future studies in stroke patients, whether imaging or clinical, should consider results according to lesion location.

This study has several limitations. First, the MT modalities differed from those used in clinical practice. Usually, individuals are asked to accompany the visualized movement by trying to move the impaired hand. This condition was not applicable to healthy individuals, as the intention would have resulted in a movement that would have masked the brain activation related to the visual feedback. Although this modality without movement intention can be applied to the patient, it does not appear to be optimal. A magnetoencephalography study in healthy individuals showed that contralateral M1 activation was greater when feedback was associated with the intention to perform the movement [15]. Another limitation concerns the task. We analysed a simple task (hand opening/closing), However, a study conducted in healthy individuals and people after stroke showed that MT-related brain activation was greater when the task was complex [55]. It would have been interesting to explore differences between simple and complex motor tasks.

Moreover, our study has some recruitment-related limitations. This study was carried out only on healthy subjects, whereas it investigates rehabilitation methods (i.e. MT and VT) used in the rehabilitation of stroke patients. While this was an important step, it will be necessary to evaluate these activation patterns and any clinical differences between the two techniques in stroke patients in future studies. In addition, we only recruited right-handed individuals, and the motor tasks were performed on the right side with left visual feedback. Therefore, our results cannot be extrapolated to the use of MT for the non-dominant side or to left-handed individuals. An electroencephalogram study of 13 healthy individuals found increased intracortical contralateral inhibition to movement and activation of mirror neurons [46]. However, the authors showed that the former effect was greater when participants moved their right (dominant) hand, and the latter was greater when the feedback was to the right (thus left-handed motor skills). To our knowledge, the specificity of left-handedness has not been evaluated in MT, but it is likely that different brain mechanisms are involved. Indeed, it has been shown that left-handed individuals have more bilateral activation patterns during the execution of motor tasks [56]. These different parameters should be considered in future studies.

Another limitation regarding the sample is that we only recruited young subjects, whereas some studies have shown that cortical activation patterns during motor task performance are different in older subjects. For example, one study showed that activation was more bilateral in older participants during a hand rehabilitation exercise using a multisensory glove [57]. Therefore, our results cannot be directly generalized to older adults or people after stroke, who are usually older than our study participants. Further studies in older subjects are thus warranted.

Finally, the fNIRS device did not allow us to cover the whole cortex. Therefore, we selected regions of interest (PC, PMC, and M1). However, other zones are activated during MT, such as the supplementary motor area, parietal and occipital cortices [34]. Unfortunately, this is a limitation of the fNIRS technique [12, 16]. In addition, some regions are not accessible by fNIRS, such as the cerebellum, which appears to be involved in MT [58].