The review concluded that a general bimanual network was activated when healthy subjects underwent different types of BMCG, which consisted of the bilateral M1, SMA, cerebellum, and S1. Secondly, previous studies showed that increased unimpaired hemisphere involvement and decreased positive neural coupling between the ipsilateral SMA and the ipsilateral M1 were the major differences in neural patterns between patients with stroke and healthy individuals who performed BMCG. Thirdly, excitatory NIBS applied over the left PMd and the bilateral cerebellum led to a positive influence on BMCG in healthy individuals, whereas applied over the bilateral M1 and the left IPC, they did not induce a positive effect on BMCG in healthy subjects. Finally, stroke patients with mild impairment benefited from excitatory stimulation of the ipsilateral M1, and those with severe impairment benefited from stimulation of the contralateral PMd in terms of improvement in BMIG. In contrast, chronic stroke patients with mild to moderate impairment did not benefit from excitatory stimulation of the contralateral M1 in terms of improvement in BMCG.
Regarding the findings of the general bimanual network, a study examining the effects of NIBS on healthy individuals to improve BMCG showed that excitatory stimulation over the cerebellum resulted in positive changes in bimanual coordination [41]. This aligns with our findings that all patterns activate the cerebellum during BMCG (Fig. 3), suggesting that the anterior and posterior cerebellar vermis are involved in spatial and temporal control during complex bimanual tasks. Additionally, Dietz et al. identified a neural network related to BMCG, primarily involving bilateral M1, SMA, and cerebellum [46]. However, our results indicate significant extra-activation on the bilateral S1 compared to Dietz et al.’s findings [46]. More precisely, sensory information, such as proprioceptive and cutaneous sensory input, predominantly contributes to motor learning and control. Moreover, functional goal-oriented tasks (i.e., reaching out and grasping a ball) elicit higher activation in S1 than non-functional movements [47–49]. This implies that bilateral S1 plays a pivotal role in BMCG, especially in activities demanding purposeful movements and the integration of sensory input to enhance coordinated actions involving both upper limbs or hands.
Our hierarchical cluster analysis revealed three distinct activation patterns within the brain regions when various BMCG tasks were performed. Notably, Cluster I was only associated with the execution of an object manipulation task performed while participants observed a computer screen. Object manipulation necessitates intricate fine motor control and the concurrent operation of two sensory systems—namely, proprioception and somatic sensation. Signals originating from tactile and proprioceptive fibers, which innervate the skin and muscles of the hand, are conveyed to the S1 through the ventro-postero-lateral nucleus of the thalamus and the cuneate nucleus in the brainstem. [50]. This observation is consistent with our results that the IPC in the bilateral hemispheres was more significantly activated in Cluster I than in Cluster II. In addition, visual information regarding the cursor’s location helped foster a proper motion plan. Specialised circuits in the primary visual cortex located in the occipital lobe can aid in achieving this visuomotor transformation through the posterior parietal cortex, PMd, and PMv. Therefore, our results indicated that the BMCG in Cluster I led to the significant activation of the bilateral MOC, right PMv, and left PMd compared to Cluster II and III [50, 51].
The complexity of BMCG varied between Clusters II and III, specifically in terms of relative phase, temporal aspects (relative interlimb frequency), and spatial elements (direction and amplitude). In contrast to the tasks in Cluster II, those in Cluster III exhibited increased complexity across all these domains. Two studies investigating age-related changes in the brain highlighted that complex bimanual motor tasks led to heightened activities in the PFC and primary M1 in both hemispheres [21, 52]. Debaere et al. proposed a positive association between spatiotemporal complexity of bilateral movements and activation in the SPC, SMA, and thalamus in both hemispheres. Additionally, they suggested that the movement frequency of bimanual tasks were positively associated mainly with activities in the bilateral CMC, PMC, and thalamus [53]. This partly aligns with our findings that complex bilateral movements activate the bilateral M1, PMC, thalamus, CMC, and PFC (Fig. 3.). However, in contrast to the findings of Debaere et al., our results demonstrated that the bimanual movements of different complexities led to significant activation of the bilateral SMA. Notably, augmented activities in the SPC in the bilateral hemispheres were observed during relatively simple bilateral tasks but not during relatively complex bilateral tasks; this outcome may be related to the neural mechanism of imitation [54].
The primary distinction between Cluster II and Clusters I and III lies in the level of familiarity that healthy subjects have with BMCG. Specifically, when healthy participants were learning a new bilateral task, bilateral M1 and SMA exhibited the most pronounced activation and the most extensive network of functional connectivity [55, 56]. In addition, some studies indicated that the cerebellum plays a crucial role in the motor learning network by facilitating SMA to M1 communication, which means that the communication between the cerebellum and motor cortical areas is vital for motor learning [57, 58]. These findings support our discovery of a cortical-striato-cerebellar network involved in motor learning, specifically bilateral M1, SMA, and the left cerebellum, which undergoes modulation changes when healthy individuals attempt an innovative bilateral task. However, we also observed increased activation in the right MTC when healthy individuals performed a novel BMCG task, which may be associated with the neural mechanisms underlying explicit and implicit motor learning processes. Previous research has shown that improved activation in the MTC is associated with increased cognitive demand during the learning and execution of new tasks [59, 60].
The key contrast observed between BMCG and BMIG in healthy individuals during a hand activity with a specific speed ratio was the enhanced activation in the right STC. Specifically, when engaging in a hand activity with a given fixed rhythm, there were generally greater demands in spatial and temporal aspects compared to activities without rhythm. Previous studies have highlighted the dominant role of the right hemisphere for spatial representation and attention, particularly in the STC [61, 62]. Heightened activation of the right STC may contribute to the proficient execution of BMCG by enabling constant monitoring of the spatial positions of both hands in healthy individuals. Our observation aligns with a study on the impact of NIBS, supporting the notion that stimulating the right STC actively enhances bimanual coordination performance, thereby promoting awareness of the spatial locations of both hands [63]. When examining task-oriented activities, it became evident that the disparity between BMCG and BMIG in healthy individuals resulted in excessive activation of bilateral S2. This can be attributed to the fact that S2 is responsible for merging sensory input from both hands, an essential requirement for coordinating movements between the two sides during cooperative actions rather than non-cooperative ones [35, 64]. Interestingly, our findings are in agreement with the research conducted by Michels et al., which also revealed similar outcomes, indicating that younger individuals exhibit heightened activation in bilateral S2 when performing task-specific BMCG as opposed to BMIG [65]. These findings support our assertion that bilateral S2 is critical in task-based BMCG in healthy individuals.
The primary neural distinctions observed between stroke patients and healthy individuals performing BMCG included heightened engagement of the unaffected hemisphere and reduced positive neural connectivity between the ipsilateral SMA and the ipsilateral M1. Neural coupling during BMCG is achieved through the exchange of information from each hand to both the hemispheres through the corpus callosum while producing effective FC between specific sensorimotor areas, resulting in the appropriate control of bimanual coordination [35]. Compared with healthy individuals, stroke patients who received nerve stimulation of the affected hand when performing BMCG demonstrated increased engagement of the unaffected hemisphere, which may involve S2 activation. The S2 acts as an integration area for sensory input from the right hand, left hand, or both hands, and it is significantly activated after stimulation [22, 66]. In addition, the bilateral S2 is more activated in BMCG (i.e., bottle opening and closing) than in non-cooperative movements [65]. In other words, increased bilateral S2 activation aids in performing BMCG in patients with stroke compared with that in healthy individuals. Another finding of this review is that an imbalance in inter-hemispheric inhibition arises from diminished neural coupling between the ipsilateral M1 and the contralateral SMA and between bilateral SMA, disrupting the neural functions and leading to their overactivation. Balanced interhemispheric coupling is crucial for the coordination of bilateral patterns [67]. This imbalance in inter-hemispheric inhibition results in impaired coordination. In addition, underactivation in ipsilateral M1 induced by decreased positive neural coupling between ipsilateral M1 and ipsilateral SMA that negatively affected BMCG in stroke patients closely aligned with our results of excitatory NIBS applied over the ipsilateral M1 improved BMCG in stroke patients with mild impairment.
We observed differences among the effects of NIBS on healthy individuals for improving BMCG. This difference was due to the distinction in the stimulation sites, outcome measures, and ceiling effect among the included studies. First, excitatory stimulation over the cerebellum induced positive change in bimanual coordination because the anterior and posterior cerebellar vermis perform spatial and temporal control during complex bimanual tasks [24]; this is consistent with our result that all types of activation patterns activate the cerebellum during BMCG. Moreover, virtual lesions over the right STC disrupted bimanual coordination performance because the right STC is mainly responsible for spatial awareness and exploration associated with both object- and spatial-related information, which aid in fostering the spatial location of the two hands [63]. Second, compared with sham stimulation, excitatory stimulation over the bilateral M1 did not trigger a positive effect in terms of bilateral task performance (i.e., error rate and movement time). This may have been due to the stimulation frequency used; specifically, beta may not be the optimal stimulation frequency for motor learning compared with the other frequencies (e.g., gamma and alpha) [68]. This may be due to beta-tACS increasing beta in the M1 by entrain endogenous beta rhythms and causing cortical reorganisation associated with motor learning. Third, healthy individuals aged 18–30 years have great bimanual coordination, which may have led to a ceiling effect (where the effects of 10- and 20-Hz tACS were indistinguishable), leaving only little room for significant enhancement after tACS. Similar findings were found in some other studies on healthy individuals [69, 70].
The effects of NIBS applied in the M1 for improving BMCG are highly dependent on the impairment levels of patients with stroke. This is consistent with the bimodal recovery model - mild impairment needs a re-establishment of IHI, while severe impairment needs contralesional compensation [52]. Specifically, stimulation over the M1 on the impaired hemisphere than on the unimpaired hemisphere led to significant improvement in bimanual coordination in stroke patients with mild impairment. This is because, compared with other impairment levels, stimulation over the ipsilateral M1 leads to a great response, which is positively associated with motor recovery [71, 72], which is also consistent with our findings that reduced positive neural coupling between ipsilateral M1 and ipsilateral SMA was observed when performed BMCG in stroke patients compared with healthy individuals (Table S2). However, stroke patients with severe impairment did not benefit from simulation over the ipsilateral M1 because strong transcallosal inhibition from the unaffected hemisphere to the affected hemisphere reduced patient response to NIBS [28]. They may benefit from stimulation over the contralateral PMd. This is because the contralateral PMd compensatorily supports the general motor functions in severe stroke patients and can trigger more facilitatory and less inhibitory effects on the ipsilateral M1 than in those with other impairment levels [73, 74]. Thus, we guess the improvement of BMCG in stroke is because of the improvement of general motor impairment rather than a specific effect on BMCG. The difference in effect between individuals with healthy and stroke also supports this statement that excitatory stimulation over the M1 is ineffective, whereas, over the PMC and cerebellum, it is effective in healthy individuals.