Nineteen healthy volunteers participated in Experiment 1, where a movement lasting 200 ms was presented [8 male, average age 28 ± 1.7 years, 18/19 right-handed (Oldfield, 1971), all naïve to the protocol].

Seventeen healthy volunteers were recruited for Experiment 2, where a movement lasting 720 ms was shown [9 male, average age 27.8 ± 2.2 years, 16/17 right-handed (Oldfield, 1971), all naïve to the protocol].

All subjects had normal or corrected-to-normal vision and were medication-free in the week prior to the session. Subjects were seated comfortably on a chair with their left forearm resting on a table and both hands covered from view, while observing a PC monitor 1 m in front of them, on which was displayed a third's person left hand (Figure 1A). In each session comprising 4 sets, we presented 136 repetitions of the same left index-finger movement for observation. Repetitions were separated each other by a transitional Blank lasting 1–2 s. Twenty repetitions of the same hand not moving (“Still”) were also presented (in a ratio of 1:7 of the other stimuli), and in such cases the absence of movement was cued by the word “Still” in the preceding transitional Blank. We also included 4 Stills which were un-cued. Subjects were requested to notify the assistant when no movement was observed after the black-screen if not warned as “Still”; if the subject did not report the un-cued Still, data from the whole set was discarded as sign of non-attentiveness, and repeated (this occurred for 7 subjects, but only in 1 set). Un-cued Stills were present in Experiments 1 and 2, with the cued “Still” only in Experiment 1.

For each repetition one TMS pulse was delivered, its timing relative to the observed movement was systematically but randomly explored (see below). The whole range of stimuli was presented in 4 sets, with 3 min rest between sets. The stimuli were prepared/assembled using Pinnacle System Software: The hand was filmed, and a single slow finger movement recorded. From the 25 fps recording, several frames were selected: 1 Still frame (the last frame preceding finger movement during recording) and 9 Movement frames at different finger positions. To create the Movement video stimulus, frames were arranged in this order: 3–4 s × Still frame + Movement frames + 3.3–3.8 s Still frame (Figure 2). To create the Still video all frames were repetitions of the Still frame (Figure 2A). The same frames were use to create the 200 ms (Figure 2B) and 720 ms (Figure 2C) movement. While the 720 ms movement frames sequence included 9 frames showing different finger positions during the extension phase recorded (≈16% MAX, ≈33% MAX,…. ≈66% MAX,… 100%MAX) and then reversed to show a symmetrical movement; the 200 ms 5 movement frame sequence was (≈33% MAX, ≈66% MAX, 100%MAX, ≈66% MAX, ≈33% MAX) (Figure 2); thus both movements were identical in amplitude and symmetry, but of different duration.

TMS-pulses were delivered using a monophasic Magstim 200² stimulator (Magstim Company, Whiteland, Dyfed, UK) with 70 mm figure-of-8 coil. The coil was oriented tangentially to the skull, with the handle pointing 45° back/downwards to induce currents in the postero-anterior direction (Figure 1B). The stimulation intensity was 130% of individual rest motor threshold (RMT), calculated on the hot-spot of the first dorsal interosseous (FDI) on the right hemisphere, which was marked. RMT was the minimum intensity required to produce 50% of responses of ~50 μV in 10 consecutive trials (Rossini et al., 1994), while looking at a black screen and with the muscle at rest. Stimulation whilst watching the movies was applied over the hot-spot of the FDI on the right hemisphere, one pulse for each video stimulus. Timing of TMS-pulses relative to the observed movement was controlled by the addition of a 2 × 5 cm black rectangle in the Still frames (Figure 2), turning to white in the frame appearing 80 ms prior to movement-onset, and at the same time in the Still video stimuli. This region was obscured on the monitor observed by the participants but was covered by a photocell attached to the PC screen and sent to the CED1401 as the “video-trigger.” TMS was delivered at the following delays relative to the movement-onset: 10, 120, 220, 360, 500, 900, and 1500 ms (Figure 2). Delays were randomized across the experiment, (Signal Software; CED, UK) and the same intervals were used for control Still video stimuli.

The SPSS (SPSS Inc) software version 15.0 was used to analyze the data. The analyzed variables were the same for all experiments:

Normality of the distributions was checked by Kolgómorov-Smirnov tests for one sample. Univariate analyses of variance with repeated measures (ANOVA_RM) were used in all experiments. It evaluated changes as a function of the observed action and time-course from movement onset. The Mauchly test was used to evaluate sphericity; if violated, the Greenhouse coefficients (Ԑ) were employed to correct the degrees of freedom. The Bonferroni's correction was applied for multiple pair-wise comparisons. Significance set at p ≤ 0.05.

Experiment 1 specifically evaluated the modulation of the CSE at different delays. We used an ANOVA_RM with two within subject factors: DELAY had 7 levels (Control, 120 ms… 1500); and MUSCLE with 3 levels (FDI, THU, and EXT).

The analysis of the CSE rebound-effect included a condition not dependent on delay from movement-onset. It was evaluated by an independent ANOVA_RM, with factor CONDITION (500 ms, Still, and 1500 ms) and factor MUSCLE (as before).

Before conducting the analysis we normalized the data. Then, for each muscle, we calculated the mean value at the Control delay across all subjects, which served as the divisor for the rest of the delays for that give muscle.

Subsequently, Experiment 2 included a group of subjects observing movements lasting 720 ms. We normalized the data (as before, average of Control responses) and analyzed the difference in modulation of CSE in the two groups, inspecting DELAY and MUSCLE and a between-subjects factor GROUP (levels G200, G720).

For Experiment 3 the ANOVA_RM was similar to Experiment 2 (DELAY × MUSCLE × GROUP), but DELAY included only Control and 500 ms delays, and GROUP the TMS-group and CMS-group. The amplitudes of the MEPs were normalized (TMS and CMS are not directly comparable). For each muscle and group we calculated the average of the response in the Black condition, which became the divisor for the Control and 500 ms delays in the corresponding muscle and group.

In Experiment 4, we analyzed the responses at Control and 500 ms delays in the three muscles, and the effect of cTBS. The ANOVA included DELAY, MUSCLE, and TIME-POINT evaluation (before and after cTBS). We also statistically evaluated the effect of cTBS on Black and Still responses in the same way, but excluding the factor DELAY. Before conducting the ANOVA, data was normalized; for each muscle we calculated the average of all subjects' responses before and after cTBS in the Black condition, which served as divisor for all values in the corresponding muscle.

In the first experiment subjects passively observed repetitions of the movement lasting 200 ms. Interleaved at random between the sequences with movements they also observed trials showing the same hand but not moving (Still), which was cued in advance of the presentation. The TMS pulses were delivered at specific delays from movement-onset, and the Control delay was delivered before MO but when it was also expected to be observed. The Still repetitions served to evaluate any rebound effect on CSE modulation during MO.

Figure 3A shows overlaid traces of the CSE exploration at the different delays in a representative subject. CSE was significantly modified across the different delays during MO [F_{(6, 108)} = 5.860; p < 0.001]. Figure 3B shows this pooled-muscle effect, which was not significantly different for the three muscles evaluated [F_{(12, 216)} = 0.966_ε=0.428; p = 0.444]. This is shown in Figure 3C which illustrates the overlapping of the responses of the three muscles. The three muscles showed a significantly reduced CSE (reduced MEPs) at 360 and 500 ms after movement-onset (Bonferroni p = 0.043 and p = 0.014, Figure 3B).

Subsequently we asked if the effect in the three muscles was larger than that obtained during the Still condition, which therefore had no effect due to MO expectancy. For this, we compared the magnitude of the effect at 500 ms and at the longer delay of 1500 ms, vs. the cued-Still hand, showing that there was a significant difference [F_{(2, 36)} = 4.338; p = 0.021]. Post-hoc testing indicated a significantly reduced CSE at 500 ms compared to Still (Bonferroni p = 0.026), while at 1500 ms the CSE had recovered to similar levels to Still (Bonferroni p = 0.887), Figure 3D.

These results show an active reduction of the CSE in the three muscles examined during MO, which is more than can be explained by the mere removal of MO expectancy, and is present for observation of a movement lasting 200 ms.

In the second experiment we asked if this effect was dependent on the duration of the movement observed. For this, we compared the modulation of the CSE at the different delays during the observation of a 200 ms movement, to the profile resulting from observing a movement of the same amplitude and symmetry (extension and retraction), but of 720 ms duration. Again, the CSE was significantly modulated at specific delays during MO [F_{(6, 204)} = 12.278_ε=0.647; p < 0.001] and the effect was not differentially expressed in the three muscles [F_{(12, 408)} = 1.319_ε=0.496; p = 0.251]; remarkably these effects were not significantly different for the groups observing a movement lasting 200 or 720 ms [F_{(12, 408)} = 0.855_ε=0.496; p = 0.529]. This can be observed in Figure 3E, which shows 200 vs. 720 ms responses at each delay.

While Experiments 1 and 2 explain a reduction of the CSE during the observation of movements, the TMS technique does not allow us to localize the origin of such an effect. In Experiment 3 we asked if this effect could be explained by the modulation of the spinal cord circuits, for which we used electric CMS; or by the modulation of intracortical circuits, explored by paired-pulse TMS. During these experiments subjects observed repetitions of the movement lasting 720 ms, and the stimulation was delivered at the Control and 500 ms delays. We also included several randomized videos showing a Blank (Black), which served to normalize data and allow the comparison of the different techniques (TMS and CMS).

Figure 4A shows that the CSE evaluated with single pulse TMS is significantly reduced at 500 ms after movement-onset [F_{(1, 11)} = 12.955; p = 0.004], and this effect was not differentially observed in the three muscles [F_{(2, 22)} = 0.209; p = 0.813]. It is clear from Figure 4A that the modulation of the SE (using CMS) was significantly different to that observed with TMS [F_{(1, 17)} = 6.324; p = 0.022]; also the modulation of the SE was not significantly different at 500 ms compared to Control [F_{(1, 6)} = 0.342; p = 0.580, Figure 4A], and this was not differentially expressed for the three muscles [F_{(2, 12)} = 0.006; p = 0.992]. These results therefore exclude a role of SE in the reduction of CSE observed at 500 ms during MO.

Subsequently, we asked if an increase in SICI at 500 ms could be responsible for such an effect, but SICI was not significantly modified at 500 ms compared to Control [F_{(1, 11)} = 0.925; p = 0.357, Figure 4B]. Finally, instead of a reduction of the ICF at 500 ms compared to Control, which might explain the drop in CSE, the circuits explored showed a significant increase in excitability [F_{(1, 11)} = 5.004; p = 0.047, Figure 4C]. The effects seen were not significantly different in the three muscles under study, both for SICI [F_{(2, 22)} = 1.215; p = 0.316, Figure 4D purple traces], or ICF [F_{(2, 22)} = 0.309_ε=0.667; p = 0.651, Figure 4D red traces].

Thus, far we have observed a significant reduction of the CSE at 500 ms during MO which is not explained by changes in the excitability of the spinal motorneurons. The effect was also not explained by a modulation of the excitability in the inhibitory circuits of M1, as explored by SICI (circuits generating I₂–I₃ descending activity). However, conversely, we observed an increase in cortical excitability of the circuits investigated by ICF. In Experiment 4 we explored the specific modulation of cortical circuits generating I₁ descending volleys at the time of increasing ICF (500 ms after movement-onset), suggesting it to be responsible for the net decrease in CSE. The subjects observed the videos while single TMS pulses were delivered (PRE) at Control and 500 ms delays and, in addition, during Black and Still videos (all randomized in order). After this, 40 s of cTBS was applied to M1 to inhibit the intracortical circuits generating I₁-descending activity, and single pulse TMS was delivered (POST), as before.

Figure 5A shows a significant effect of cTBS on modulating cortical output [F_{(1, 10)} = 11.996; p = 0.006], expressed similarly in all three muscles [F_{(2, 20)} = 0.584; p = 0.567, Figure 5B], and obtained during the observation of the Blank. Remarkably, the CSE was also significantly reduced by the cTBS during the observation of a Still hand [F_{(1, 10)} = 17.331; p = 0.002, Figure 5C], and the three muscles did not show any significant different responses to this effect [F_{(2, 20)} = 0.061; p = 0.941, Figure 5D]. A representative example on the hot-spot, FDI, during Black is shown in Figure 5E.

If we focus on the effects during MO at the two different delays from movement-onset, we also observed a significant modulation of the excitability induced by cTBS [F_{(1, 10)} = 5.982; p = 0.03] and remarkably, the effect was significantly different at Control and 500 ms delays [F_{(1, 10)} = 5.361; p = 0.043, Figure 6A]; but not seen in a significantly different manner for the three muscles [F_{(2, 20)} = 0.421; p = 0.662, Figure 6B]. Therefore, the CSE at PRE (before cTBS) was significantly different between Control and 500 ms (Post-hoc test p = 0.007) in agreement with Experiments 1–3, but not at POST, after cTBS (Post-hoc p = 0.873, Figure 6A). This effect is explained because the CSE at the Control delay was significantly reduced by cTBS (Post-hoc test p = 0.009), whereas at 500 ms the CSE was not significantly modified by cTBS (Post-hoc test p = 0.155, Figure 6A). Figure 6C shows a representative example (one subject) of the recordings obtained with stimulation at the hot-spot.

In each the sets of results reported above (Experiments 1–4) the MEPs were acquired in conditions where the levels of EMG background activity were not significantly different between delays and/or conditions.

The process we describe is triggered by a very simple intransitive movement during plain observation. No motor act was made or planned (Gangitano et al., 2001, 2004) as this is known to modulate the activity of the motor system (Decety et al., 1997). However, plain observation with no movement performance might involve a motor plan to stay still. In our experiments subjects were told to avoid any movement at the time of observation. This “conscious suppression” could have had an influence on the results, given that the mirror neuron network is involved in the inhibition or release (regulation) of intentional or automatic imitation (Brass et al., 2005; Bien et al., 2009).

Observation of movements of different complexity might also influence the modulation of the CSE of the observer. Thus, complex or/and forceful movements are known to induce greater facilitation of M1 (Alaerts et al., 2010), also affecting spinal excitability (Baldissera et al., 2001; Patuzzo et al., 2003; Borroni et al., 2005, 2008; Montagna et al., 2005; Borroni and Baldissera, 2008), which might override the stereotyped inhibition described herein.

It also is conceivable that the effect we observed at 500 ms starts earlier (present at 360 ms also for inhibition in our study). In line with this idea is the fact that mirror facilitation is observed about 200 ms after onset of simple intransitive movements, if MNS structures have been primed by observation in the 1st person perspective (Lepage et al., 2010; Catmur et al., 2011). Thus, perspective might differentially modulate the excitability of the motor system during MO (Maeda et al., 2002; Schutz-Bosbach et al., 2006; Alaerts et al., 2009a).

Our results support different roles for several cortical circuits during MO. Critically, it is known that the application of TMS recruits different cortico-spinal circuits depending on pulse intensities, direction of the induced currents, pulse-wave forms or coil type (Di Lazzaro et al., 2008), and the effect of stimulation parameters on the modulation of CSE during MO is known (Loporto et al., 2013). This might explain differences between published studies.

In conclusion, during human MO there is a phase of stereotyped modulation in the excitability of cortico-cortical circuits. This permits an early MNS facilitation in M1, but which is suppressed also at cortical level, perhaps aiming at avoiding overt motor replication. At the same time, the spinal circuits maintain the level of excitability induced by MO expectancy, which might be boosted if the observer wishes or requires respond to the action.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.