Twenty-four healthy volunteers (Experiment 1: 9 men and 3 women, mean age 25.7 ± 5.6 years; Experiment 2: 7 men and 5 women, mean age 29.2 ± 5.2 years) were enrolled in this study after providing written, informed consent. All subjects were self-reported right-handers. Baseline characteristics of subjects are shown in Table 1.

The present study was based on the global guidelines for care in the use of TMS (Rossi et al., 2009). In the first stage of recruitment, all subjects filled out a questionnaire designed to exclude those with contraindications; however, none reported neurological impairment or contraindications to TMS. The study was approved by the local ethics committee at the Nagasaki University Graduate School of Biomedical and Health Sciences. All experimental procedures were conducted in accordance with the Declaration of Helsinki (World Medical Association, 2013).

Subjects were seated on a reclining chair 80 cm away from a PC monitor (RDT234WX-Z, MITSUBISHI, Japan, 23-inch, resolution, 1920 × 1080 pixels; refresh frequency 60 Hz) and instructed to keep both hands in a pronated position on a horizontal board attached to the chair’s armrest. They were instructed to keep the right forearm as still and relaxed as possible while paying attention to the visual stimuli presented on the PC monitor. To ensure passive observation of the video clips, the experimenter’s only instruction to the subjects was “You should stay alert while observing a hand,” before starting the experiment.

We shot two types of video for the experimental task: one was rapid and the other was slow. The definition of a rapid or slow task was whether or not the elements of the movements could be recognized at a normal playback speed. We shot the task movies with an electrode placed on the actor’s skin and set the timing of the TMS trigger by means of electromyogram (EMG) data during task execution.

In order to trigger stimulation at specific times, 210 individual frames were converted to JPEG files and shown in succession to obtain the animation effect. At normal speed, the presentation time of each frame was twice the length of the refresh interval used by the PC monitor (refresh interval = 16.67 ms). The differences in video speed determined the presentation time of each frame (normal speed 33.3 ms/frame; slow speed 66.67 ms/frame; fast speed 16.67 ms/frame). The timing TMS trigger was established for each specific file: it occurred at the same point in the action on the video—for example, when the actors hand was opened fully. Previous studies have found that MEP amplitude correlates significantly with finger aperture during grasping movements and peaks when finger aperture is maximal (Gangitano et al., 2001, 2004). Moreover, previous studies have also revealed a clear correspondence between the time course of the modulation of MEPs during action observation and the temporal pattern of EMG activity (Borroni et al., 2005; Montagna et al., 2005). We decided the timing of the TMS trigger based on these reports. The review written by Naish et al. (2014) showed that “the timing and extent of muscle activity during execution needs to be confirmed by EMG recording, rather than estimated based on knowledge of the primary functions of muscles”. Therefore, we recorded EMG data and assessed the subjects’ muscle activity during task execution.

We used Lab Chart 8 (AD Instruments, Australia) to combine the task movement recordings and the EMG data, and to establish the timing of the TMS trigger. The EMG data were analyzed using the root mean square (RMS) of every 33.3 ms, and the decision on the TMS timing was made by checking the JPEG file and the EMG data as described above. TMS was delivered twice in both experiments. In Experiment 1, we established TMS delivery times as in our previous study (Moriuchi et al., 2014). We decided the trigger point labeled “open” (80/210 frames) was the time the model’s hand had opened to the widest extent just before catching the ball and “catch” (85/210 frames) was the time the model had just caught the ball. By contrast, in Experiment 2, we established the TMS trigger point labeled “open” (54/210 frames) was the time at which the model’s hand aperture was the widest and “lift” (119/210 frames) was the time the model had just lifted the ball.

Prior to the action observation task, M1 baseline excitability at rest was assessed in each subject by recording 20 MEPs while the subject observed a white cross on a black screen under controlled conditions. Subsequently, the experimenter instructed the subject to watch the video without any additional mental effort. All subjects observed the same video clip played at three different speeds (slow, normal and fast). TMS was delivered once for each video clip, randomly at “open” or “catch” (Experiment 1) and “open” or “lift” (Experiment 2). For each subject each speed condition was used in 20 trials. Ten trials used the “open” phase of TMS delivery and ten used the “catch” (Experiment 1) or “lift” (Experiment 2) phase. MEP amplitudes were calculated for each TMS delivery.

We used a computerized pulse-generation system (LabView, National Instruments, Austin, TX, USA). To ensure that TMS was always delivered at the correct time and that the experimental design was correctly implemented, the speed order (slow, normal and fast) was randomized by the experimenter, and the order of TMS delivery times (“open” or “catch”/“open” or “lift”) was randomized by the LabView system. Each of the six possible trial conditions (three speeds × two phases) was replicated 10 times for each subject, for a total of 20 trials per speed, and a further 20 trials were control trials involving TMS and MEP data collection at intervals of 10 s during a sham trial, giving a total of 80 trials per subject.

Surface EMG activity was recorded in the first dorsal interosseous (FDI) and in the right abductor digiti minimi (ADM), using pairs of 9 mm diameter Ag-AgCl surface cup electrodes (SDC112, GE Healthcare, Japan). Surface EMG signals were amplified and filtered at a bandwidth of 5–3000 Hz using a digital signal processor (Neuropack sigma MEB-5504, Nihon Kohden, Japan), and were transferred to a computer for off-line analysis, using an A/D converter (PowerLab16/30, AD Instruments, Bella Vista, NSW, Australia).

At the beginning of the experiment, we identified the optimal TMS coil position for evoking MEPs in both the right FDI and the right ADM (the hotspot). TMS was delivered to the left M1 hotspot, marked with a pen on a swimming cap covering the scalp of each subject. TMS employed a 70 mm figure-of-eight coil connected to a magnetic stimulator (Magstim 200, Magstim, UK). The coil was placed tangentially to the scalp with its handle pointing backward and rotated approximately 45° away from the mid-sagittal line. Care was taken to maintain the same coil position relative to the scalp throughout the experiment. The resting motor threshold (MT) was defined as the lowest stimulus intensity that evoked an MEP at least 50 μV in amplitude in the right FDI and in the ADM in five out of 10 trials. The test stimulus intensity was set at 110%–130% of the resting MT and produced a control MEP with a success rate that ranged from 35% to 55%, mean 46.6 ± 5.4%, for Experiment 1; 33% to 51%, mean 44.2 ± 6.6% for Experiment 2. The mean size of the control MEP for the FDI and ADM was approximately 0.5–1.0 mV. Throughout the experiments, subjects were instructed to avoid inadvertent movements that could give rise to background EMG activity. For each muscle in each trial, the 20 ms period preceding TMS triggering was checked for background EMG activity.

If background EMG data was found, data from both muscles in the trial were rejected. MEP amplitude (peak-to-peak) was measured over each muscle in every trial. MEP amplitude was analyzed using peak-to-peak values and expressed as a percentage of the mean amplitude under control conditions. The data were statistically analyzed using a one-way analysis of variance (ANOVA) with the factor “video speed” (slow, normal and fast) to investigate whether the MEP amplitude was modulated compared with the control condition. Moreover, the data were analyzed statistically using a three-way analysis ANOVA with the factors “video speed” (slow, normal and fast), “muscle” (FDI, ADM), “timing” (open, catch (Experiment 1); open, lift (Experiment 2)). We employed Tukey’s post hoc test for multiple comparisons for further analyses. In all analyses, the p level for statistical significance was set at p < 0.05. All analyses were performed using statistical analysis software (SPSS version 22.0, IBM, Armonk, NY, USA).

Some studies have shown that M1 excitability is enhanced by the social features that characterize an observed action (Donne et al., 2011; Bucchioni et al., 2013; Sartori et al., 2013). These studies suggest that M1 excitability is modulated by the participant’s understanding of the social intention and context implied by the observed action. In our study, the experimental action video clearly presented someone catching a ball thrown by someone else in front of him (Experiment 1), or someone reaching and lifting toward a ball (Experiment 2). Therefore, it was possible to understand both the context and intention. These studies would predict that M1 excitability would be enhanced by viewing the video in this context, which indeed occurred. In the current study, we identified that MEP amplitude was found to differ significantly among speeds solely in the rapid movement task. MEP amplitude was not modulated among speeds in the slow movement task. MEP amplitude was higher when subjects observed an action played at a relatively slow speed than at normal or fast speeds. We suggest that these results demonstrate that the effect on M1 excitability of manipulating the speed at which the action is viewed during passive observation differs, depending on the type of movement that is being observed.

Two previous studies of action observation have investigated changes in movement velocity and in the speed at which video of a recorded movement was played. The latter of these, the “video speed study” is most similar to the present study. The former study—on movement velocity itself—identified a significant correlation between M1 excitability and the velocity of the observed upper limb movement: the faster the observed movement, the greater the increase in M1 excitability (Agosta et al., 2016). Further, another “movement velocity study” recording EMG during performing 1.0 Hz to 1.6 Hz flexion extensions of the right wrist revealed that the EMG data of actual movement at 1.6 Hz is significantly higher than at 1.0 Hz (Borroni et al., 2005). The difference in EMG activity might also have affected the modulation of M1 excitability while observing different actual movement velocities.

On the other hand, there are similar “video speed studies” in addition to the present study (Wrightson et al., 2016). In Wrightson’s study, M1 excitability was not modulated between the video speed conditions. The present study also was not modulated on the fast video speed condition in either Experiment 1 and 2. From the results of these “video speed studies”, they revealed that M1 excitability was not modulated during observation of a video playing at a faster than normal speed.

When we compared the movement velocity and video speed studies, we found significant conflicting results between the studies and their characteristics. The “video speed study” found no differences in muscle activity in the video because only the playing speed of the original movie was changed. On the other hand, in the “movement velocity study”, as velocity of movement increased, the amplitude and area of the EMG bursts increased (Mustard and Lee, 1987), therefore, the muscle activity was different among the movement velocity conditions. This difference suggests there might be a different mechanism, rather than just changing movement velocity, that modulates M1.

The present study adopted two types of observational movement task velocity (slow movement and rapid movement). M1 excitability was modulated when observing a rapid movement at slow speed. Moreover, the ADM was significantly facilitated compared to the FDI when a rapid movement was presented at a slow video speed. Everyone expects to be able to recognize elements of rapid movement played at a slow speed that would not be visible to the naked eye at normal or fast speeds. On the other hand, we would not acquire new information such as kinematic movement cues from observing movement at a slow speed. Previous studies using fMRI have found that the PMv and IPL constituent parts of the AON are particularly involved in the subject’s ability to understand motor-related components of observed actions. They suggest that AON activation depends on whether movement elements such as kinematic cues are recognized or not (Ogawa and Inui, 2012). Moreover, the inferior frontal gyrus (IFG), which is strongly connected to M1, reflects the kinematic features of observed actions (Grafton and Hamilton, 2007).

In light of these previous studies, the present study hypothesized that subjects would recognize the elements of the action more easily (and therefore, the AON would be activated) when the replay speed of a video clip of rapid movement was reduced, and the modulation of M1 excitability would be accompanied by AON activation, particularly the PMv and IPL. In contrast, the slow speed movement component of the present study and the arm crank ergometers task of a previous study suggest that the AON was not dependent on the speed of the task video, and that the subject would recognize movement elements without changing the video speed. Therefore, modulation of M1 excitability was not demonstrated under those conditions.

In the present study, video speed modulated M1 excitability only in the ADM but not in the FDI. Previous studies revealed that the muscle-specific modulations of M1 excitability were highly specific to the type of grasping observed. In a whole hand grasp condition (i.e., opposition of the thumb with all fingers), the ADM was facilitated more strongly than the FDI. On the other hand, the FDI was facilitated more strongly than the ADM in a precision grip condition (i.e., the opposition of the thumb with the index finger; Koch et al., 2010; Sartori et al., 2012; de Beukelaar et al., 2016). This visuo-motor transformation involves the anterior intraparietal area (AIP)-PMv-M1 network (Jeannerod et al., 1995). The AIP provides the PMv with the object’s grasp-related information such as its size and shape. Then, PMv outputs the information to the M1 representation of selected muscles involved in the task (Davare et al., 2010). Since the present study adopted the whole hand grasp as an experimental task, the ADM was preferentially activated, in a muscle-specific manner, compared to the FDI. For this reason, the ADM might have been affected by changes in video speed.

An investigation of the effects of changes in the static position of the shoulder on the cortico-spinal excitability of hand muscle revealed that static shoulder position affects the modulation of cortico-spinal excitability only in ADM and not in FDI. This suggests that the FDI is controlled independently without any restrictions of proximal-distal synergies (Dominici et al., 2005). Further, research into the effect of voluntary teeth clenching (VTC), on cortical inhibition revealed that MEPs with posterior-lateral current direction (preferentially elicited I3-wave) in FDI were significantly decreased by VTC when the hand muscle was slightly contracted; however, MEPs in ADM were not altered by VTC. This finding suggests that the FDI muscle may be finely controlled with less restriction of VTC (Takahashi et al., 2006).

In the present study, to summarize, the FDI muscle may be activated independently without being influenced by other factors. This might be the reason why the FDI was not affected by changing video speeds. Our findings showed that there is a possibility that the impact of video speed is dependent on muscle properties. However, since this is still speculative, more studies are needed into the effect of video speed on the FDI and other muscles.

The present study did not find that M1 excitability was significantly modulated during action observation compared to a control condition in either experiment condition. However, similar results have been found in previous studies (Sakamoto et al., 2009; Ohno et al., 2011).

A previous study suggested that measurements of corticospinal excitability by TMS during action observation may be an excellent paradigm for probing the AON (Maeda et al., 2002). However, the method used in that study could only probe the AON indirectly, and could not explore the functional roles of other cortical areas in which mirror neurons have been found, such as the PMv and IPL. Several current studies aim to directly explore the individual effects of various neural networks involved in the AON during action observation by means of the bifocal twin-coil TMS method. This is a conditioning-test TMS paradigm in which a test stimulus is applied to M1 after different delays and after a conditioning stimulus is delivered to another cortical area (Davare et al., 2008, 2009; Koch et al., 2010; de Beukelaar et al., 2016). In particular, studies have investigated the excitability of the connections linking PMv and M1 and identified that excitability was not changed at rest, but facilitated in a muscle-specific way during action observation (Koch et al., 2010; de Beukelaar et al., 2016). Activation of the AON during action observation induces specific neurophysiological changes in some of the cortico-cortical connections of the human motor system, such as the PMv-M1. This reinforces the idea that increased excitability of the motor cortex during action observation is mediated by AON activation.

Other methods use transcranial direct current stimulation (tDCS). Subjects underwent cathodal tDCS to the area of the AON, and M1 excitability during action observation was significantly reduced following cathodal stimulation (Enticott et al., 2012). It would be possible to search for AON activation more directly by combining tDCS, adjusted for the excitability of AON, and in the next case assess the excitability of M1 by using the single-pulse TMS method. On the other hand, there is value in using fMRI as a methodology that can search for activation across the entire brain. However, it would be necessary to consider the experimental protocol if fMRI were to be used, because the length of exposure is different for different video playing speeds.

If we wish to discover more about the impact of different video speeds on AON in the future, we will need to research the use of other imaging protocol such as a tDCS and fMRI protocol.