Previous Transcranial Magnetic Stimulation (TMS) studies have shown that the observer's motor system is facilitated by the sole observation of motor actions. However, it has not been possible so far to decide whether the observer's motor system resonates primarily with the observed movement direction or the observed muscle activity, as both factors usually co-varied in these action observation studies. Here, we applied TMS to the wrist extensor and flexor during the observation of wrist motions such that the posture of the observer and the model in the video were either congruent or incongruent. Due to this manipulation, it was possible to disentangle whether the observer's primary motor cortex (M1) is facilitated in accordance to either the observed movement direction or the observed muscle activation. Findings revealed that M1 resonated predominantly according to muscle-specific rather than direction-specific parameters of observed movements. More specifically, muscle-specific facilitation was maximal during congruent postures and remained evident, even though to a lower extent, during incongruent postures in which muscle activation and movement direction parameters were discordant. Our findings support the hypothesis that M1 contributes to action observation, by representing the observed movement in intrinsic, muscle-related coordinates. This transformation from extrinsic to intrinsic coordinates might be an important prerequisite for action understanding and imitation. Additionally, our data offer a neurophysiological explanation for interference that emerges when an action is performed while an incongruent action is observed.