The presentation of weak noise effectively enhances the detection of stimuli in the sensory systems and perceptual processes (Collins et al. 1996, 1997; Wells et al. 2005). Our results showed that subthreshold somatosensory information obtained using electrical stimulation reduced the motor error from the target value in the visuomotor tracking task. Subthreshold electrical noise induced stochastic resonance and improved grip accuracy, but substantial inter-individual differences were observed. Notably, participants who exhibited large motor errors during the visuomotor tracking task were considerably affected by somatosensory noise in the sensorimotor system.

The sensory system processes information from inside and outside the body, providing information for motor control via the central nervous system. Stochastic resonance has been hypothesized to affect the motor system by facilitating the sensory system. Electrical stimulation of the vestibular system has been used to improve postural balance in the elderly (Fujimoto et al. 2016). In the somatosensory system, the presentation of subthreshold somatosensory information improves the ability to move pegs and square boxes, as well as the maximum precision grip force in patients with upper limb motor dysfunction (Seo et al. 2014; Nobusako et al. 2019). However, improvements in motor function due to somatosensory noise have been reported in the performance of all motor tasks, and how the stochastic resonance effect extends to various types of muscle movements is unclear. We investigated whether somatosensory information improves motor accuracy by using a visuomotor tracking task and demonstrated that the noise reduces motor errors in various sub-periods in the FG, FR and Constant phases.

Subthreshold electrical stimulation was previously shown to reduce force fluctuations during low-level, steady force exertion of the lower muscles (Kouzaki et al. 2012). Our study using grip movement revealed the same result, with the coefficient of variation being considerably small during the Constant phase in the ST0.9 condition. Additionally, although our study demonstrated a significant reduction in the absolute motor error during the FG phase, the difference in the FR phase was insignificant. These results indicate that subthreshold electrical stimulation enhanced the ability to slowly increase grip force but not to slowly decrease it; however, the stochastic resonance effect varied across different muscle contraction modalities. Afferent activity from muscle spindles differ depending on the contraction style (Burke et al. 1978; Al-falahe et al., 1990). In addition, a neuroimaging study comparing FG and FR in humans showed lower activation of the primary motor cortex during force relaxation (Spraker et al. 2009). Similarly, somatosensory information processing in the primary somatosensory cortex differed between FG and FR (Wasaka et al. 2012). These findings indicate that the activity of the cortical sensorimotor areas differs depending on the muscle contraction pattern. In the present study, the intensity of the electrical stimulus presented during the visuomotor tracking task was constant, but stochastic resonance by sensory noise seemed to have different effects depending on the muscle contraction modes.

Noise enhances sensorimotor integration processes involved in motor regulation. Since somatosensory information plays a crucial role in motor accuracy and force regulation (Rothwell et al. 1982), it is conceivable that the facilitation of the somatosensory system by stochastic resonance affects motor control ability. Variability in the discharge rates of motor units and the relationship between agonist and antagonist activities causes motor errors (Vallbo and Wessberg 1993; Laidlaw et al. 2000). Our finding that noise reduces motor errors suggests that somatosensory information slightly weaker than the sensory threshold affects these underlying mechanisms. In the present study, electrical stimulation was delivered to the median nerve that innervates the first lumbrical, second lumbrical, opponens pollicis, flexor pollicis brevis and abductor pollicis brevis muscles. Therefore, this stimulation is assumed to affect sensorimotor integration between the somatosensory and motor systems in the finger muscles innervated by the median nerve. In the tracking task similar to our motor task, the alpha-gamma linkage increases the afferent discharges from muscle spindle (Hulliger and Vallbo 1979). The possible explanation for reduction of motor errors might be that subthreshold electrical stimulation caused excitation in Ia fibers from the muscle spindle, inducing stochastic resonance and affecting the alpha-gamma linkage. Another possible explanation of our finding is that electrical stimulation caused excitation of Ib fibers from the Golgi tendon organs or cutaneous information acted on the motor system (Cordo et al. 1996; Collins et al. 2005). At intensities below or just above the sensory threshold, the cutaneous nerves appear to be excited, but because Ia and Ib fibres have similar diameters, it is unclear whether electrical stimulation produces differences in their activation. Therefore, the region of the central nervous system involved in promoting sensorimotor function in response to noise has not been clarified. During a tactile discrimination task using fingertips, discrimination ability is improved when noise below the sensory threshold is presented to the fingertips; however, discrimination ability is improved even when noise is presented to the dorsum of the hand or thumb, which is not related to discrimination (Lakshminarayanan et al. 2015). Although our results showed that presenting noise to the nerve innervating the moving muscles improved motor performance, further investigation is required to elucidate the effects of presenting tactile information to other body parts related to the movement.

Somatosensory information is transmitted to the spinal cord and cerebral cortex. The stimulus intensity of the sensory threshold evoked neural activity in the primary somatosensory cortex (Lin et al. 2003), and presenting continuous noise of somatosensory information below the sensory threshold enhanced the amplitude of somatosensory-evoked potentials (Seo et al. 2015). Additionally, when stochastic resonance by sensory noise improves motor accuracy, corticomuscular coherence increases (Trenado et al. 2014). These findings suggest that feedback of somatosensory information from the muscles and skin exists owing to grip movement, and a neural mechanism that reduces motor errors by noise possibly exists in the sensorimotor system.

As electrical stimulation is applied to the wrist, the influence of attention on tactile information is assumed to improve the ability to adjust the grip force of the fingers. However, improvement in the grip force adjustment control was more pronounced with the subthreshold intensity of the ST0.9 condition than that of the NO condition. Under these conditions, the participants were unable to judge whether electrical stimulation was being presented during the motor task. Thus, factors other than the influence of attention on electrical stimulation acting on the sensorimotor system improved the grip force adjustment ability.

Previous studies investigating motor performance using stochastic resonance have shown that subthreshold somatosensory noise is effective (Kouzaki et al. 2012; Mendez-Balbuena et al. 2012; Germer et al. 2019); however, the intensity above the sensory threshold has not been investigated. Since the ST1.1 condition did not considerably improve motor performance, our results suggest that a higher intensity of noise does not enhance motor adjustment ability. Because the median nerve contains both motor and sensory nerves, stronger stimulation intensity increases sensory nerve activity and simultaneously excites motor fibers. Motor nerve activity may have caused recurrent inhibition in motor neurons (Hultborn et al. 1971), and motor errors may not have changed in the ST1.1 condition. Given that the mean sensory threshold in all subjects was 50.7% of the motor threshold, the intensity in the ST1.1 condition induced no excitation of motor fibers, thereby exerting minimal influence on recurrent inhibition. Although we found that stimulus intensity lower than the sensory threshold is important to induce stochastic resonance for enhancing motor ability, it remains unclear whether this intensity is optimal.

A significant correlation indicated that the effect of weak electrical stimulation was more for participants with larger absolute motor errors in the NO condition across all contraction phases. However, no significant differences in absolute motor errors were observed between the NO and ST0.9 conditions in the FR phase, although it decreased in the ST0.9 condition. Based on these results, we suggest the discrepancy between results of ANOVA and correlation analysis is attribute to the large variations in the amount of reduction during the FR phase. In addition, the correlation analysis showed that weak electrical stimulation had a large stochastic resonance effect on the participants with large motor errors in the NO condition but had little effect on those with small motor errors. Establishing a method to enhance the motor function of those with small motor errors in the NO condition is important. Further studies to resolve the issue of individual differences by improving the noise presentation method are warranted.

Appropriate sensory information processing and sensorimotor integration in the central nervous system enables the coordination of motor control, resulting in postural maintenance, motor accuracy, and force coordination. We found that the weak sensory information reduced motor errors in various styles of muscle contraction. In addition, this phenomenon showed characteristic changes in the sub-periods before and after changing the contraction styles and when performing the same contraction style stably and continuously. Our research is an experimental verification in a limited setting involving switching between muscle contraction types and continuing a stable contraction. Although it is unclear from our results whether the subthreshold sensory noise influenced on grip force control based on the visual information of exerted force or the ability to accurately control muscle position, our findings demonstrate that stochastic resonance can be applied for precise motor control in humans. Given that our research results are expected to be applied to improving daily movements using fingers as well as specialized motor skills, they have important applications in improving rehabilitation and the development of physical abilities.