The results show that the direction of vibration-induced postural shifts is a function of the selected stimulation location around the torso, while the magnitude of the postural shifts is a function of the tactor type used to generate the vibrotactile stimulus. A significant postural shift towards the location of the applied vibration was observed when stimulation was applied over the internal oblique and erector spinae muscle locations. The direction of the observed postural shift was not dependent on tactor type. These findings suggest that cutaneous information from the skin over the muscles of the torso contributes to the proprioceptive internal representation of the upper body and its orientation. Indeed, the directional shift was congruent with a postural response resulting from the lengthening of an abdominal muscle which is accompanied by skin stretch. Such responses also occur when vibration stimulates the muscle spindles [36, 37]. Hence, the vibration-induced activity of cutaneous receptors is likely interpreted as a skin stretch corresponding to proprioceptive information, as shown for distal joints . The latency of the postural response for stimulation over the internal oblique and erector spinae locations is substantially greater than that of a reflex response, which is known to be less than 100 ms [38, 39]; thus, a significant role of reflex contribution and muscle proprioception to changes in posture is ruled out. This hypothesis, discussed in detail in our previous study , is briefly outlined here.
Vedel and Roll  and Ribot-Ciscar et al.  have shown that mechanoreceptors are very sensitive to mechanical vibration with stimulations in the range of 200–500 μm peak-to-peak displacement. The magnitude of both the postural shifts and the RMS sway occurring in response to an applied vibratory stimulus was significantly larger when the C2 tactor was employed versus the Tactaid tactor. This was to be expected, since the stimulation magnitude was approximately four times greater for the C2 tactors than for the Tactaid tactors. Furthermore, the subjects reported that the magnitude of the vibration across the locations was perceived to be the same for a given tactor type. Although it could not be experimentally controlled, given the subjective responses of the participants we assumed that the tactor contact pressure was fairly equally distributed around the torso by the elastic belt. However, the subjects indicated that the perceived vibration intensity (i.e., displacement amplitude) was greater for the C2 tactors than for the Tactaid tactors. This difference in perception is in agreement with the difference in postural responses and is well correlated with vibration strength. Furthermore, this finding is in agreement with investigations by Martin et al. , who show that the strength of vibration-induced proprioceptive activity increases with the magnitude of the vibration stimulus. Kavounoudias et al.  and Wierzbicka et al.  have also shown that the postural responses induced by vibration of the ankle muscles increase with stimulation magnitude. Therefore, we assume that due to the greater strength of the C2 tactor, a larger number of tactile receptors are recruited by C2 than Tactaid stimulation, which in turn increases the associated compensatory response. The efficiency of the stimulation may also be greater for linear tactors, such as the C2, than for inertial actuators, such as the Tactaid, since the C2 may produce a larger deformation of the skin due to the unique direction of travel of the generated pulse waves. Hence, a better efficiency may be obtained by a more secure driving of the cutaneous receptors. In other words, the consistency of receptor response to each vibration cycle would be greater for normal stretch than for shear stretch.
The drifts of postural responses are monotonous and reach a peak at approximately the same time for both tactor types; however the peak is greater for the C2 tactors than for the Tactaid tactors. One possible interpretation is that more secure driving and a larger recruitment of tactile receptors increase response speed, as indicated by the results related to vibration-induced illusions, since the speed of vibration induced illusory movements [37, 43, 44] or real movements  is in proportion to frequency. Furthermore, the average value of the PSD mean power frequency (frequencies less than 0.6 Hz for both A/P and M/L directions) was not significantly different in the presence or absence of vibrotactile stimulation regardless of the tactor type or tactor location. Since the measured postural sway frequency lies within the normal range of less than 1.0 Hz [13, 45], vibrotactile stimulation does not appear to induce a disruptive increase in sway frequency, but rather an adjustment of posture associated with proprioceptive information.
Vibration applied to the skin over the external oblique muscle locations did not induce a significant shift regardless of the tactor type. Indeed, postural stability is usually greater in the M/L than A/P direction during normal stance  and, in the present study, the hip-width separation of the feet also contributed to a high lateral stability. Hence, a small vibration-induced change in sensory information is less likely to induce a compensatory postural response in the direction corresponding to the action of these muscles, since stability may not be perceived to be compromised.
The results of the present study show a vibration-induced inclination of the torso; however, the measurements of postural trajectories at the torso level do not allow for the description of a possible reorganization of posture implicating a multi-segmental response (e.g., head, upper body, lower body). Thus, further investigation is necessary to assess the relative contribution among different body segments (i.e., the reorganization of different body segments for postural coordination).
Our experimental findings suggest that tactor type and application locations should be carefully considered when designing vibrotactile displays to be used around the torso. Moreover, the choice of instructions concerning corrective movements requires additional investigation to determine their compatibility with the non-volitional response to the vibrotactile stimulation. It has yet to be determined whether or not the use of attractive instructional cues (“move in the direction of the vibrotactile stimulus”) facilitates a postural response during vibrotactile biofeedback balance applications. The instructional cue may change the cognitive interpretation of the cutaneous information generated by the vibration and thus the compatibility of the response direction with that stimulation.