Axis-specific neuro-musculoskeletal contributions to postural stability and vibration avoidance in mice

35 36 In modern industrial societies, animals including humans are 37 frequently exposed to external mechanical oscillations such as whole-body 38 (WB) vibration. To ensure postural stability and effective move ment, 39 animals must appropriately regulate their responses to such vib ratory 40 disturbances. Therefore, understanding how animals respond to v ibration 41 is of growing importance. However, compared with humans, knowledge of 42 mouse responses to WB vibration remains limited, as previous st udies 43 have primarily examined anesthetized or euthanized mice. Here, we used 44 high-speed video-based quantitat ive analyses to characterize bo dy part 45 movements in both awake and anesthetized mice under vertical, 46 longitudinal, and lateral WB vibration. Awake mice exhibited lo wer 47 resonance frequencies and smaller displacement amplitudes than 48 anesthetized mice during both vertical and lateral WB vibration , whereas 49 longitudinal WB vibration produ ced minimal differences. These r esults 50 indicate that the active neuro-musculoskeletal (NMS) system contributes to 51 postural stability in a vibration axis-dependent manner. Axis-s pecific 52 vibration control may reflect an evolutionarily acquired balanc e between 53 postural stability and locomotor efficiency. Vibration modeling f u r t h e r 54 suggests that the NMS system acts as a vibration absorber by re ducing 55 stiffness and providing feedback-controlled active damping, analogous to 56 active vehicle suspensions. Behavioral preference tests revealed that mice 57 selectively avoid vertical WB vibration at specific frequencies , but not 58 lateral or longitudinal vibration, regardless of resonance freq uencies. 59 These findings suggest that resonated body part displacement du ring WB 60 vibration do not directly determine vibration discomfort, and t hat vertical 61 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 3 WB vibration along the gravitational axis exerts a distinct imp act on 62 postural control and behavior in mice. This study provides new insights 63 into the biomechanics and behavior of quadrupeds exposed to WB 64 vibration, with analogies to four -wheeled vehicles offering a useful 65 perspective. 66 67 68 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 4 Main manuscript 69 70 Animals are exposed to various kinds of mechanical vibration, w hich can be 71 produced by external factors such as other animals (e.g., pups carried by mother mice), 72 and internal factors such as self-movement (e.g., head oscillat ions during running). To 73 maintain stable posture and generate appropriate movements, body parts must respond to 74 such vibratory disturbances in a controlled manner 1,2. In humans, whole-body (WB) 75 vibration has been extensively investigated because of its effe cts on responses, 76 biodynamics, health, and discomfort 3–6. 77 By contrast, much less is known about how non-human animals res pond under 78 WB vibration 7,8. Previous studies in rodents ha ve often measured resonance properties 79 in anesthetized or euthanized animals 9,10. This was due to the difficulty of maintaining 80 animals in a fixed position wit hout stress during vibration. Although such approaches 81 yielded valuable basic data on body transmissibility, anesthetized (or euthanized) animals 82 inherently lack active neural and muscular control, thereby overlooking the essential role 83 of the neuro-musculoskeletal (NMS) system in responding to vibr ation and stabilizing 84 posture 1,2,11. Consequently, the dynamics of W B vibration responses in awake animals 85 remain poorly understood. This h as been a critical limitation, because the neuro-86 musculoskeletal (NMS) system m ay function as a vibration suppre ssor like a spring-87 damper system 3–5 and precise biomechanical understanding of animal’s body 88 transmissibility during WB vibra tion requires data from awake animals. In addition, 89 previous studies using animals was focused mainly on vertical W B vibration, not lateral 90 nor longitudinal one, and measurement points of body transmissibility were very limited 91 (usually a single body point) 9,10, due to technical difficulties. These limited and 92 insufficient knowledge on vibration responses in animals remains to be complemented. 93 Mice represent a powerful animal model for addressing this issue. As quadrupeds, 94 their postural and locomotor strategies can be compared with th ose of other mammals, 95 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 5 while their small size, and pharm acological and behavioral trac tability enable precise 96 experimental manipulation and diverse measurements with many ex perimental tool kits 97 available. Moreover, recent developments of high-speed image ac quisition and image 98 processing techniques make it possible to simultaneously monitor movements at multiple 99 body parts of a mouse during WB vibration. 100 Here, we investigated the contribution of the NMS system to WB vibration 101 responses along three axes in mice. Using high-speed video-based analysis, we examined 102 multiple body part responses during vertical, longitudinal, and lateral WB vibration in 103 awake and anesthetized mice. Comparison of those data revealed axis-specific 104 contribution of the NMS system during WB vibration, which is reminiscent of vibration 105 control strategy in four-wheeled v ehicles. To examine vibration preference in mice, we 106 utilized a two-shuttle-box behavioral test and found a novel vertical axis-specific 107 vibration avoidance in mice. 108 109 In the present study, both awake and anesthetized mice were placed in custom-110 made chambers (mouse boxes) and were exposed to whole-body (WB) vibrations along 111 three axes (vertical [up-down], longitudinal [fore-aft], and la teral [left-right]). The 112 waveforms of the vibrations were sinusoidal with varying frequencies (3 to 40 Hz) and a 113 fixed peak-to-peak displacement amplitude of 1 mm, unless other wise noticed (Figs. 1-114 3). The relatively small size of the mouse boxes allowed awake mice to maintain a typical 115 hunched resting posture, which was advantageous for obtaining reliable measurements of 116 body part movements in response to WB vibrations under a consis tent posture 12 117 (Supplementary Videos 1 and 2). For anesthetized mice, the feet were affixed directly to 118 the vibrating table to maintain a natural hunched posture as much as possible 119 (Supplementary Fig. S5). Movements of various body parts and references in response to 120 WB vibrations were measured by tracking predefined points marked on the skin and the 121 mouse box prior to the experiments (see Methods). The measured displacement 122 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 6 amplitudes of the reference points ( ~1 mm, consistent with the intended stimulation 123 amplitude) confirmed that the vibration stimuli were well controlled under all conditions, 124 except at the low frequencies of 3–5 Hz (Supplementary Fig. S6; see Methods). The 125 awake state represents a natural physiological state with intac t neural control, whereas 126 the anesthetized state corresponds to a state lacking neural co ntrol of body posture 13. 127 Comparing awake and anesthetized mice enabled us to investigate the functional 128 contribution of neural control to the musculoskeletal system. Therefore, we quantitatively 129 analyzed the parameters of the measured movements in various bo dy parts, such as 130 displacement amplitude and phase difference relative to the ref erence movement (i.e., 131 input vibration) to examine diff erences between the awake and a nesthetized states. The 132 frequency at which vibration occurs most readily and is amplified in an object, is referred 133 to as the resonance frequency 8. We also analyzed the resonance frequencies of multiple 134 body parts in both awake and anesthetized mice during three-axis WB vibrations (vertical, 135 longitudinal, and lateral). 136 137 Body part movements during vertical WB vibrations 138 Figure 1A depicts the experimental setting and the body part po sitions used for 139 displacement measurements during vertical WB vibrations, with the symbol code also 140 applied to Figure 1B, 1D, and 1E. Figure 1B shows pooled data of the peak-to-peak 141 displacement amplitudes in awake and anesthetized mice. In awak e mice, the relative 142 displacement amplitudes of the m iddle-to-posterior body parts (‘middle’, ‘back’, ‘lower 143 back’, and ‘tail base’) along th e vibration axis were noticeabl y enhanced within the 144 frequency range of 10-16 Hz (Fig. 1B, left panel; green and blue symbols compared with 145 a black dashed line), indicative of resonance. In contrast, ant erior body parts (‘shoulder’ 146 and ‘head’; Fig. 1B, left panel, magenta symbols) exhibited min imal or no enhancement 147 of the displacement amplitude in this frequency range, and damp ing effects emerged at 148 lower frequencies than the other regions except ‘heel’. Notably, the relative displacement 149 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 7 of the ‘head’ remained constant and close to the input vibratio n amplitude (black dased 150 line) at less than 15 Hz, indica ting no clear resonance. At fre quencies below 10 Hz, 151 displacement amplitudes of all body parts were similar to the r eference (i.e., relative 152 amplitude close to 1), although the amplitudes of lower body parts located near the basal 153 vibrating surface tended to be smaller. In anesthetized mice (F ig. 1B, right), the relative 154 displacement amplitudes were substantially larger than those in awake mice across all 155 body parts except for the ‘lower b ack’, ‘tail base’, and ‘heel’, the three regions located 156 closest to the vibrating surface, suggesting reduced damping in the absence of active 157 neuromuscular control. 158 Figure 1C1 shows the resonance frequencies, defined as the frequencies at which 159 the relative displacement amplitude reached its maximum for eac h body part, based on 160 the averaged frequency–amplitude profiles shown in Figure 1B. In all body parts, awake 161 mice exhibited substantially low er resonance frequencies than a nesthetized mice. This 162 difference remained consistent and statistically significant ev en when resonance 163 frequencies were measured from individual frequency-amplitude profiles for each mouse 164 and then averaged by body part (Supplementary Fig. S2A, left; anesthetized mice [n = 6] 165 vs. awake mice [n = 5-7], P < 0.05). In the analysis of maximal relative displacement 166 amplitude, awake mice showed significantly smaller amplitudes at anterior-to-mid body 167 regions, specifically the ‘shoulder’ and ‘middle’, than anesthetized mice, regardless of 168 whether the data were derived from group-averaged (Fig. 1C 2) or individual profiles 169 (Supplementary Fig. S2A, right). 170 At frequencies of 20 Hz or higher, awake mice exhibited a profound decrease in 171 displacement amplitudes (i.e. damping effect) across all the bo dy parts, in clear contrast 172 to anesthetized mice (Fig. 1B). Notably, in awake mice, the anterior body parts (‘shoulder’ 173 and ‘head’) and the ‘heel’ began t o exhibit a damping effect at a lower frequency ( ~15 174 Hz) than the posterior body parts (~20 Hz; ‘middle’, ‘back’, ‘lower back’, and ‘tail base’). 175 In contrast, anesthetized mice showed only minor damping effects starting around 25 Hz, 176 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 8 with a slight decrease in amplitude observed across all body parts at 30 Hz (Fig. 1B, right). 177 At frequencies above 30 Hz, quan tification of displacement ampl itude and phase 178 difference in anesthetized mice became unreliable due to highly irregular waveforms (e.g., 179 double-peak complex waveforms at the ‘lower back’ in Supplementary Fig. S3A and B), 180 considerable inter-animal variability, and deviation from the s imple sinusoidal-like 181 patterns observed in the ‘reference’ (Supplementary Fig. S3). One characteristic complex 182 displacement waveform exhibited a relatively small positive-goi ng hump during the 183 downward phase of displacement, as seen at the 'tail base' and 'shoulder' in (A) and (B) 184 of Supplementary Fig. S3. Another distinct pattern, observed at the 'tail base,' 'lower back,' 185 and 'middle' in (C) of Supplementary Fig. S3, consisted of the fusion of two positive peaks, 186 leading to a broader and larger displacement waveform with an increased apparent cycle 187 time compared to that of the 'reference' waveform (i.e., input vibration cycle time). These 188 complex waveform patterns suggest that the body parts of anesth etized mice 189 mechanically behave as coupled oscillators, where interconnected body segments contain 190 two or more oscillators (i.e., a multi-degree-of-freedom system ) that interact during 191 vertical WB vibration. Some anest hetized mice exhibited clear a mplitude damping at 192 certain body parts (e.g., the 's houlder' in Supplementary Fig. S3A and C, 8 mice), while 193 others showed little damping effect (Supplementary Fig. S3B, 3 mice). Together, these 194

indicate that the active NMS system in awake mice effec tively attenuates body 195 part movements at 20 Hz and highe r frequencies, whereas in the absence of 196 neuromuscular control, damping becomes evident only at 30 Hz and higher. Furthermore, 197 the findings suggest that the ante rior body parts, particularly the ‘head’ and ‘shoulder’, 198 are stabilized more effectively by the active NMS system during vertical WB vibration 199 than other regions. 200 Next, we analyzed the phase difference (i.e., phase delay) of b ody part 201 movements relative to the ‘reference’ movement during vertical WB vibration (Fig. 1D). 202 Compared to anesthetized mice, awake mice exhibited a greater p hase delay (negative 203 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 9 phase difference) in all body parts at much lower frequencies. In both awake and 204 anesthetized mice, phase delay emerged around the resonance fre quencies of each body 205 part (10–15 Hz in awake mice; 20–25 Hz in anesthetized mice), c onsistent with 206 predictions from simple vibration theory 14,15. Meanwhile, Figure 1E shows that in awake 207 mice, relative displacement along the longitudinal (i.e., fore- aft) axis, orthogonal to the 208 applied vertical WB vibration, was negligible compared to the v ertical displacement of 209 the ‘reference’ (i.e., input vertical displacement) (Fig. 1B). These results suggest that 210 mouse body parts may respond to vertical WB vibrations in a manner analogous to a base-211 excited, forced vibration of a single-degree-of-freedom (SDOF) mass-spring-damper 212 (MSD) system 3,4,16. Notably, the relative amplitude responses of ‘shoulder’ and ‘ head’, 213 which were efficiently damped ar ound their resonance frequencie s (10-15 Hz), became 214 smaller at higher frequencies (presumably above √2 times the natural frequency of the 215 system) than those of other, less-damped body parts (e.g., ‘middle’ and ‘lower back’) (Fig. 216 1B, left). In other words, there was no adverse (i.e., compromi sing) effect on high-217 frequency vibration isolation in the well-damped ‘shoulder’ and ‘head’ compared with 218 the less-damped ‘middle’ and ‘lower back’ (Fig. 1B, left). This response profile 219 contradicts the predictions of a simple base-excited forced vib ration model 3,14,17, and 220 instead favors an active vibration isolation model 17 (Supplementary Fig. S4; see 221 Discussion). 222 To gain mechanistic insights on how the NMS system contributes to suppression 223 of body part movements under vertical WB vibration, we made a r ough estimation of 224 stiffness k (i.e., spring constants) based on a SDOF MSD system. We assume d that the 225 measured resonance frequency clos ely approximates the natural f requency ( fn) of the 226 system. While this assumption may result in a slight underestim ation of the stiffness (k) 227 due to the presence of damping (Table 1), such a compromise is unlikely to affect the 228 relative comparison between awake and anesthetized mice. Under this assumption, k can 229 be determined by the two parameters, the mass ( m) and the resonance frequency, as 230 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 10 described in equation [2] in the Methods section and in Table 1 8,18. The two body parts, 231 the ‘middle’ and the ‘head’ was focused for their k estimation. This is because the ‘middle’ 232 was reasonably considered as the center of the total mouse body mass and was good 233 approximation to the SDOF MSD system with lumped parameters, wh ereas the ‘head’ 234 was a very important body part since it includes the brain and contains the sensory organs 235 contributing to vision (eyes), a udition (ears), and vestibular sensation (vestibules of the 236 ears). The mass of the ‘middle’ and the ‘head’ was assumed to b e 22.0 g (0.022 kg) and 237 2.8 g (0.0028 kg), respectively, based on the typical body weight (BW) of an adult mouse 238 and the mean measured head-to-BW ratio. The estimated stiffness (k) values in awake 239 mice were less than half of those in anesthetized mice (Table 1 ). This finding suggests 240 that the active NMS system may function to reduce stiffness for responding to vertical 241 WB vibration, as interpreted withi n the framework of a simple S DOF MSD vibration 242 model. 243 The NMS system can contribute to suppression of body part movements through 244 damping properties in the vibration system when the NMS system is considered 245 analogous to a base-excited SDOF system with lumped parameters: a mass (m) connected 246 to a spring (stiffness, k) and a damper (damping coefficient, c ) 3,4. For vibration control, 247 there are two control models in the SDOF-MSD vibration system: a passive control model 248 where a damping property is fixed, and an active control model where damping properties 249 are actively modulated (Supplementary Fig. S4) 17. To examine where the NMS system 250 may behave as passive or active control over external vertical WB vibration, the relative 251 displacement amplitude (i.e., vibration transmissibility) of the mouse body part ‘middle’ 252 in response to vertical WB vibration was fitted with the two control models (see Methods, 253 Supplementary Fig. S4). The ‘middle’ region was considered the most suitable for model 254 fitting due to the closest region to the mouse’s center of gravity, and the mass was fixed 255 to 22.0 g as described above. As an active control model, we employed the “skyhook” 256 control model where the active control force is applied via fee dback depending on the 257 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 11 absolute velocity of the mass. The frequency response curve of the relative displacement 258 amplitude in the ‘middle’ was not fitted well to the passive co ntrol model, whereas the 259 active control (skyhook control) model explained the measured d ata reasonably with 260 stiffness 239.78 (N/m) and csky 1.9382 (N •s/m) (Supplementary Fig. S4). These results 261 indicate that the NMS system in mice may function as the active damping control over 262 vertical vibration like an active suspension of a vehicle. 263 264 Body part movements during longitudinal WB vibrations 265 In awake mice exposed to longitudinal (fore-aft) WB vibration (Fig. 2A), relative 266 fore-aft displacements increased between 5 and 12 Hz, typically exhibiting a single peak 267 in most body parts except at the ‘tail base’ and ‘heel’ regions (Fig. 2B, left). In 268 anesthetized mice, relative displacement amplitudes showed a si milar amplification 269 between 8 and 13 Hz, also with a single peak (Fig. 2B, right). Regarding phase difference 270 in awake mice, the ‘tail base’ and ‘heel’ exhibited relatively small phase shifts, while the 271 ‘back’, which is the most distant from the vibrating floor, sho wed a markedly delayed 272 phase response (exceeding -180 ° at frequencies above 15 Hz) (Fig. 2D, left; see also 273 Methods). Resonance frequencies of body parts in awake mice tended to be slightly lower 274 than those in anesthetized mice (Fig. 2C 1 and Supplementary Fig. S2B, left). However, 275 no significant differences were observed in the maximum relative displacement amplitude 276 between the two groups (Fig. 2C 2 and Supplementary Fig. S2B, right), suggesting little 277 damping effect exerted by the NMS system during longitudinal WB vibration, except at 278 the ‘tail base’ and ‘heel’, which are physically closer to the vibrating floor. 279 Unlike vertical WB vibration, l ongitudinal (fore-aft) WB vibrat ion produced 280 markedly larger relative displacements along the vertical axis (orthogonal to the stimulus 281 direction) in awake mice within the 10-20 Hz range (Fig. 2A and 2E). These responses 282 suggest that mouse body movements during longitudinal WB vibrat ion cannot be 283 explained by a single-axis model and instead involve multiple-d egree-of-freedom 284 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 12 dynamics. In the awake state, with the body supported above the ground by four limbs, 285 longitudinal vibrations may induce rotational, seesaw-like oscillatory vertical movements 286 (pitching) around the body’s center of gravity, located near the ‘middle’ body point (Fig. 287 2A). This argument is supported by Supplementary Video 2 and by the observation that 288 the orthogonal vertical displa cements of anterior and posterior body parts, farther from 289 the ‘middle’ point, were considerably larger during fore-aft WB vibrations (Fig. 2E). 290 291 Body part movements during lateral WB vibrations 292 In contrast to vertical and longitudinal WB vibration, awake mi ce exposed to 293 lateral WB vibration (Fig. 3A) exhibited complex frequency–disp lacement response 294 profiles, with two distinct peaks at 7 Hz and 12 Hz in most bod y parts (Fig. 3B, left). 295 These multi-peaked profiles suggest that body displacements during lateral WB vibration 296 may follow a multiple-degree-of-freedom vibration model in awake mice. Notably, only 297 posterior body parts (i.e., the ‘tail base’ and ‘lower back’) s howed displacement 298 amplitudes reduced to or below the amplitude of the applied vibration stimulus (Fig. 3B, 299 left, blue symbols), indicating th at these regions are damped e ffectively and vibration-300 resistant during lateral WB vibration. Unlike under longitudinal WB vibration, orthogonal 301 fore-aft displacements were negligible across all body parts du ring lateral WB vibration 302 (Fig. 3E), similar to the results observed under vertical WB vi bration. In anesthetized 303 mice, relative displacement amplitudes were markedly greater than those in awake mice 304 across all body parts, exhibiting single response peaks at 9 Hz (Fig. 3B, right). 305 Resonance frequencies derived fr om the group-averaged displacement profiles 306 were lower in awake mice compared to anesthetized mice (Fig. 3C 1). This trend was 307 confirmed by the analysis of resonance frequencies calculated from individual frequency–308 displacement profiles (Supplementary Fig. S2C, left). Statistically significant reductions 309 in resonance frequency were observed in the anterior body parts of awake mice compared 310 to anesthetized mice, including the ‘forehead’, ‘head’, and ‘back1’ (anesthetized mice [n 311 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 13 = 6] vs. awake mice [n = 7], P < 0.01). 312 Analysis of the maximum relative displacement amplitudes along the lateral 313 vibration axis revealed that awake mice exhibited significantly reduced displacements 314 across all measured body parts, with particularly pronounced suppression observed in the 315 medial regions of the body (‘middle’ and ‘back2’; Fig. 3C 2; Supplementary Fig. S2C, 316 right). These results suggest that the NMS system suppresses body part movements during 317 lateral WB vibration, functioning as a robust damping and stabi lizing mechanism in 318 response to lateral perturbations. 319 Notably, unlike in vertical and longitudinal WB vibration conditions, most body 320 parts in awake mice exhibited phase differences (phase delay) e xceeding -180° at 321 frequencies above 15 Hz, with even greater phase delays observe d at frequencies above 322 20 Hz during lateral WB vibration (Fig. 3D, left). At 20 Hz, re lative displacement 323 amplitudes remained over 0.27 across all body parts and exceede d 0.5 at the ‘back1’, 324 ‘middle’ and ‘back2’ points (Fig. 3B, left), indicating that phase-difference measurements 325 in this frequency range were reliable. These pronounced phase d elays beyond -180° in 326 awake mice may reflect either a passive multiple-degree-of-freedom vibration system or 327 active vibration control mechanisms. 328 Although Figure 3D shows gradual increase in phase delay in awa ke mice and 329 excess of -180° phase difference at around 15 Hz, analysis base d on the steady-state 330 response at fixed vibration frequency only cannot discriminate whether the phase 331 difference is really delayed beyond -180° or advanced below -180° compared to the 332

theoretically. Thus, to investigate how this large ph ase delay may develop 333 during lateral WB vibration in awake mice, we utilized a freque ncy-modulated single-334 sweep vibration stimulus (Supplementary Fig. S7A). The frequency of the stimulus was 335 continuously and gradually changed from 10 Hz to 30 Hz at a slo w rate of 0.5 Hz/s (1-336 mm peak-to-peak displacement amplitude). Moreover, to examine distinctive responses 337 of left and right body parts, symmetrical left and right body p arts were included among 338 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 14 measurement points and displacem ent responses of body parts wer e measured from the 339 rear of an awake mouse (Supplementary Fig. S8A, see Methods). 340 Analysis of the data along the lateral vibration axis using the frequency-341 modulated lateral vibration stimu lus suggests that frequency-de pendent progressive 342 increase in phase difference is continuously developed from around -90° to over -270° in 343 upper body parts (Supplementary Fig. S8C, ‘top’ and ‘upper hip’, magenta and blue), and 344 that characteristics of left and right body parts were similar (Supplementary Fig. S8B and 345 C, filled triangles and solid lines [left] vs. open squares and dotted lines [right]). However, 346 the data analyzed along the vertical axis orthogonal to the sti mulus indicate large 347 displacement (reaching over 0.25 mm) of most body parts except for ‘top’ 348 (Supplementary Fig. S8D). This resembles the results of longitu dinal (fore-aft) WB 349 vibration experiments (Fig. 2E), implying a similar possibility that lateral WB vibration 350 may induce rotational, seesaw-like vertical oscillations between the left and right sides of 351 the body. To examine this possibility, phase difference along the vertical axis (orthogonal 352 to the stimulus axis) between symmetrical left and right body p arts were analyzed 353 (Supplementary Fig. S8E). These results support the above possi bility and revealed that 354 vertical body part movements have a constant antiphase (near -180°) relationship between 355 left and right corresponding pa rts (Supplementary Fig. S7B), ir respective of lateral 356 vibration frequencies (Supplementary Fig. S8E). 357 358 Mouse preferences for different WB vibration conditions 359 The above experiments revealed that mouse body-part responses to whole-body 360 (WB) vibration differ markedly depending on the direction of vi bration (vertical, 361 longitudinal, or lateral) and its frequency (Figs. 1-3). These characteristics are 362 summarized in terms of displacement magnitude in Table 2. The differential properties of 363 body-part movements under various vibration conditions may influence mouse behaviors, 364 particularly in terms of mouse pr eferences for WB vibration sti muli 19 and potential 365 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 15 concerns about adverse effects of resonant vibrations on mice 9. Using a two-alternative 366 shuttle box behavior paradigm (Fig . 4A), we examined mouse pref erences for various 367 WB vibration conditions (6–40 Hz) along three axes (vertical, l ongitudinal, and lateral 368 directions) that were all used i n the above experiments (Table 2). Interestingly, no 369 significant preference or dislike was observed for either longi tudinal or lateral WB 370 vibration at any tested frequency compared to the corresponding control condition 371 without vibration (Fig. 4B and 4C). A single exception was note d: mice significantly 372 avoided the lateral vibration at 40 Hz, spending only ~40% of t heir time in the vibrated 373 chamber (Fig. 4C). In contrast, vertical WB vibration elicited robust avoidance behavior 374 at 32 Hz and 40 Hz (Fig. 4B and 4-C), indicating a clear axis- and frequency-dependent 375 behavioral response. These results suggest that mice selectivel y avoid vertical WB 376 vibration at frequencies over 16 Hz. This pattern cannot be att ributed to body-part 377 resonance, because the resonance frequencies of body parts in a wake mice are 378 approximately 6–16 Hz across all vibration axes (Figs. 1C1–3C1, Supplementary Fig. S2, 379 and Table 2), a range in which no significant avoidance behavior was detected (Fig. 4C). 380 Therefore, we conclude that resonance itself does not account for vibration avoidance. 381 It appears that the higher the frequency of vertical WB vibrati on was, the more 382 severely mice avoided the vibration (Fig. 4C). However, the dis placement peak-to-peak 383 amplitude of the vertical WB vi bration was fixed at 1.0 mm, and the energy of the 384 vibration stimulus (i.e., peak acceleration) increased along with the frequency increment 385 in the experiments above. Thus, we next examined whether displacement amplitude itself 386 can affect mouse preferences for vertical WB vibration. To this end, we adjusted both the 387 displacement amplitude and frequen cy of the vibration stimuli, with their peak 388 acceleration held at a fixed rate of 5.05 m/s 2, equivalent to that of a 1-mm vibration 389 stimulus at 16 Hz, which did not induce avoidance (Fig. 4C and 4D). In this configuration 390 (fixed peak acceleration conditions), a trade-off relationship exists between displacement 391 amplitude and frequency (Fig. 4D), which allowed us to dissocia te the effects of 392 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 16 frequency and displacement ampli tude under a fixed mechanical e nergy level. Under 393 these vibration conditions, mice exhibited significant avoidanc e of vertical vibration 394 when the peak-to-peak displacement exceeded 7 mm at 5.1 and 6 H z, which are much 395 below the resonance frequency range (Fig. 1B, awake). In contrast, vibration stimuli with 396 smaller displacements and hi gher frequencies (e.g., 0.25 mm at 32 Hz) did not elicit 397 avoidance, as preference remained near chance level (Fig. 4D). Collectively, these results 398 indicate that both displacement amplitude and frequency influence the aversive response 399 to vertical WB vibration in mice. 400 401 402

403 In the present study, we analyz ed the movements of various body parts during 404 WB vibrations along three differe nt (vertical, longitudinal, an d lateral) axes, and 405 compared the response properties between awake and anesthetized mice. To our 406 knowledge, this is the first report to systematically examine resonance frequencies and 407 phase delays in various body parts in awake and anesthetized mi ce along the three axes, 408 revealing the differential contribution of the NMS system to mouse postural control along 409 the three vibration axes (Figs. 1-3). We found that awake mice exhibit lower resonance 410 frequencies in body parts than anesthetized mice during WB vibrations with all three axes 411 of excitation (Figs. 1C 1-3C1, Supplementary Fig. S2), suggesting that the NMS system 412 functions to maintain body resonance at lower frequencies. In terms of vibration control, 413 our results indicate that the NMS system suppresses body vibration by reducing stiffness 414 (k) of the mass (Table 1) and ac tive control of the damper (skyho ok damper) such as 415 vehicle suspension control, when a SDOF-MSD vibration system is considered416 （Supplementary Fig. S4). This du al mechanism—reduction of stiffn ess and dynamic 417 adjustment of damping—may enable animals to suppress resonance and maintain postural 418 stability during WB vibration. The present study may be the first to propose the functional 419 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 17 and mechanical roles of the NMS system in controlling body vibr ation from a vibration 420 engineering perspective. Using be havioral preference tests, we also examined the 421 behavioral significance of three-axis WB vibrations and found that mice selectively avoid 422 vertical WB vibrations in a frequency- and displacement-dependent manner (Fig. 4). 423 424 A downward shift of body resonance frequency and a softening ef fect by the NMS 425 system 426 The functional significance of lowering body resonance frequency by the NMS 427 system in mice remains to be eluc idated. From the perspective o f vibration isolation, a 428 downward shift in resonance frequency leads to reduction in vib ration transmission at 429 frequencies above ~√2 times the resonance frequency (more precisely, the system’s 430 natural frequency) 14. In vertical WB vibrations, an esthetized mice exhibited body 431 resonance frequencies ranging 20-25 Hz, whereas awake mice showed a lower resonance 432 frequency range of approximately 10-15 Hz (Fig. 1C1, and Supplementary Fig. S2A, left). 433 This implies that the mouse NMS system might effectively attenu ate vertical body 434 vibrations at frequencies above 15 Hz, a range that could be cr itical to mice due to 435 potential harm to the brain and internal organs, or disturbances to the visual, auditory, and 436 vestibular systems. From a clinical perspective, the reduced co ntribution of the 437 neuromuscular system in individuals such as infants, older adults, or unconscious patients 438 may result in elevated body resonance frequencies. This highlig hts the importance of 439 minimizing exposure to potentially harmful high-frequency WB vi brations during 440 medical transport in ambulances or other vehicles. Such vibrati ons may approach the 441 resonance range of these vulnerable individuals, leading to amp lified body movements 442 and potentially greater physiol ogical stress compared to health y adults with intact 443 neuromuscular control and lower body resonance frequencies. 444 So far, it remains unclear how t he NMS system reduces the stiff ness k of body 445 parts (i.e., the ‘softening’ effect) (Table 1), thereby lowerin g the body resonance 446 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 18 frequencies (Figs. 1-3, and Supplementary Fig. S2, left). Interestingly, similar phenomena 447 have been reported in rhesus monkeys 20 and human subjects exposed to higher 448 magnitudes of vibration 21–24. Griffin (1990, Chapter 8.4.2.1) suggested that this effect 449 might result from an involuntary loss of muscle tone by vibration, phasic muscle activity 450 excited by vibration, or the thi xotropic behavior of muscles 18. All of these hypotheses 451 involve muscle activity 25, which is consistent with our findings. However, our results 452 appear counterintuitive, because anesthetized mice, in which al l skeletal muscles are 453 relaxed and presumably more softened, exhibit effective higher stiffness; whereas awake 454 mice, in which some (but not all) skeletal muscles are contract ed and appear more rigid, 455 show lower effective stiffness during WB vibration when the mouse body is regarded as 456 a vibrating system. This intriguing discrepancy might be reconc iled with non-linear 457 properties of the complex posture control system including musc les, tendons, bones, 458 joints, ligaments, and the nervous system 2,18. Notably, several studies have reported that 459 muscle stiffness is reduced during and after contraction 26–29. In addition, muscle spindle 460 tuning can affect muscle stiffness 30, and joint/limb stiffness can be regulated by joint 461 receptors 31. As another possibility, the skeletal structures which are intrinsically stiffer, 462 might be dominant to support the body in anesthetized mice, so that body transmissibility 463 is mediated mainly by the skeletal system, while fully relaxed muscles behave primarily 464 as passive mass attached to the skeletal structure during WB vibration. 465 Zeeman et al. 10 reported spinal resonance frequencies of 8-10 Hz during 466 longitudinal WB vibration in anesthetized rats, comparable to o ur results (Fig. 2B, 467 anesthetized; Supplementary Fig. S2B left, anesthetized). In contrast, Rabey et al. 9 468 estimated much higher resonance frequencies (41-60 Hz in mice) during vertical WB 469 vibration in anesthetized or eu thanized rodents than those obse rved in our study. This 470 discrepancy may arise from diffe rences in resonance estimation methods and vibration 471 stimuli. In particular, Rabey et al. 9 employed frequency-modulated sweep vibration 472 stimuli at a rapid rate of 30 Hz/s, which may predominantly induce nonstationary transient 473 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 19 responses, whereas our study focused on steady-state responses of the mouse body during 474 vertical WB vibration using discrete, single-frequency vibration stimuli. 475 476 Active control of body vibration by the NMS system in mice 477 Our model fitting suggests that the NMS system functions as an active vibration 478 control system to suppress body vi bration during vertical WB vi bration in mice 479 (Supplementary Fig. S4). This interpretation is also supported by the frequency response 480 profiles of the more strongly da mped anterior body parts in awa ke mice (Fig. 1B, left). 481 Specifically, the damping sufficient to control resonance at ‘shoulder’ and ‘head’ did not 482 compromise higher frequency isolation; that is, the ‘shoulder’ and ‘head’ did not exhibit 483 larger amplitude responses than the less damped body parts (e.g ., ‘middle’ and ‘lower 484 back’) above 20 Hz. This response property is characteristic of active vibration isolation, 485 as utilized in automobile suspensions 17. In contrast, the seated human body exposed to 486 vertical vibration has been well described by passive control models (Supplementary Fig. 487 S4A, upper panel) 21,32. This discrepancy may refle ct fundamental biomechanical 488 differences: humans are bipedal with an upright rostrocaudal ax is, whereas mice are 489 quadrupedal with a prone rostrocaudal axis. Thus, it might be r easonable that mice with 490 four limbs employ a vibration control strategy analogous to four-wheeled vehicles. 491 Using the fitted parameters ( k and csky) and equation [2] of the active skyhook 492 control model (Supplementary Fig. S4), the damping ratio (also referred to as the damping 493 factor), Zsky of the ‘middle’ mouse body (the presumable body’s center of ma ss) was 494 estimated to be 0.42. This value is comparable to those reported in the Rhesus monkey 495 (~0.46) 20, the seated human body (0.475) 21, and walking humans (~0.3) 33 under vertical 496 WB vibration. These findings suggest that the damping ratio of animal bodies in response 497 to vertical vibration may be evolutionarily conserved across species, although it can vary 498 depending on the ethological context of each species. Such cons ervation might provide 499 biomechanical advantages by optimizing a trade-off relationship between stability against 500 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 20 external perturbations (e.g., vibration) and mobility efficiency (i.e., controlled instability 501 with energy cost) during locomotion. 502 Biological mechanisms underlying active skyhook vibration control in mice 503 remain to be elucidated. In the skyhook damper system, measurem ent and use of the 504 absolute mass velocity are requi red; however, in practice, it i s not possible to connect a 505 physical damper from a mass (i.e., the mouse body) to an inerti al reference 34. In mice, 506 head movement-corrected stable visual images together with vestibular signals encoding 507 body acceleration might be exploited to approximate absolute velocity. Group Ia muscle 508 spindle afferents act as muscle velocity sensors 31,35, and thus muscle spindle-mediated 509 velocity-dependent feedback control of muscle activity may provide damping of body 510 vibration 36, given that damping force is proportional to the velocity of the vibrating body 511 14. Moreover, group Ia muscle spindl e afferents and some Golgi tendon organs can 512 respond to vibration in tendons and muscles 31,35. Skin, fascia, joint capsules, and 513 ligaments are also innervated by mechanoreceptors that are sensitive to vibration 35. 514 Neural processing of these propr ioceptive information and the r esulting mechanical 515 output could implement active skyhook-like control of body vibration in mice. 516 517 Vibration axis-specific response properties of mouse body parts and its functional 518 implications 519 In vertical WB vibration, the present study revealed that one of the anterior body 520 parts, the‘head’ is vibration resistant and stabilized most effectively by the active NMS 521 system among mouse body parts (Fig. 1B, awake). A similar effect has also been reported 522 in standing humans 37,38. This stabilization may contribute to the stable and accurate 523 acquisition of sensory information, because most sensory organs including eyes (vision), 524 ears (hearing), vestibular apparatus (balance), nose (olfaction), and whiskers (touch), are 525 concentrated in the head where stability is critical for detect ing and processing these 526 sensory stimuli in mice. Moreover, head stability may protect the brain from mechanical 527 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 21 damage during locomotion which inevitably involves vertical WB vibration. 528 In anesthetized mice where the NMS system is inhibited, only ve rtical WB 529 vibration caused complex. coupl ed oscillation-like displacement patterns of body parts 530 (Supplementary Fig. S3), whereas no such patterns were observed during longitudinal or 531 lateral WB vibration. This phenom enon may result from the effect of gravity along the 532 vertical axis. In awake mice, the NMS system-mediated vibration control may suppress 533 such coupled oscillations, which could otherwise be injurious t o certain body parts, and 534 instead allow the body to respond a s a s i m p l e S D O F v i b r a t i o n s ystem (Fig. 1, 535 Supplementary Fig. S4), thereby minimizing mechanical stress am ong body parts. 536 Limiting the system’s degrees of freedom may also be advantageous by reducing the cost 537 of body control 39. 538 The present study found that the mid-body regions, ‘middle’ and ‘back’ move 539 maximally among the body parts, r egardless of vibration axis or the state of the NMS 540 system (active or inactive) (Figs. 1-3 and Supplementary Fig. S2). This phenomenon may 541 reflect that the mid body, including internal organs, lacks rib support, making it more 542 susceptible to vibration. 543 Curiously, one of the posterior body parts, the ‘lower back’ in awake mice tended 544 to exhibit larger vibration responses than in anesthetized mice lacking the NMS system-545 mediated control during vertical W B vibration (Fig. 1B, blue circles and C 2; 546 Supplementary Fig. S2A, right). This finding is in shar contrast to our general observation 547 that the NMS system suppresses vibration responses of body part s in awake mice (Figs. 548 1-3) and was not observed during longitudinal (fore-aft) or lat eral WB vibration (Figs. 549 2B, 2C2, 3B, and 3C2; Supplementary Fig. S2B and C, r ight). These results suggest that 550 vertical axis-specific response enhancement at the ‘lower back’ might contribute to active 551 postural control against gravity by the NMS system. During vert ical WB vibration, the 552 NMS system might functionally couple the posterior body parts ( ‘lower 553 back’, ’back’, ’middle’, and ’tail base’) to reduce mechanical stress within the posterior 554 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 22 body area. 555 Among the three axes of WB vibra tion, two distinct response pro perties were 556 observed only during longitudinal WB vibration in mice. First, vibration responses were 557 largely uniform across body parts, except for lower body region s close to the vibrating 558 floor, in both awake and anesthe tized mice (Fig. 2B). This indi cates that, unlike in the 559 vertical and lateral axes (Figs 1B and 3B), mouse body parts are equally mobile along the 560 longitudinal axis. Second, displacement responses differed only slightly between awake 561 and anesthetized mice (Fig. 2B and 2C2), suggesting that the NMS system exerts the least 562 control over body posture along th e longitudinal (fore-aft) axi s. In other words, mouse 563 body parts remain easily movable in the longitudinal direction even when the NMS 564 system is active. Analogous to four-wheeled vehicles, such mobility may be advantageous 565 for quadrupeds like mice, which primarily locomote in the fore-aft direction. In addition, 566 interestingly, mouse stride frequency during wheel running typi cally ranges from 4 to 9 567 Hz 40 ,which corresponds well to our observed resonance frequency ra nge during 568 longitudinal WB vibration in mice (Fig. 2B and 2C 1, awake; Supplementary Fig. S2B, 569 left). This correspondence may also contribute to efficient locomotion in mice. 570 Regarding the lateral (left-right) axis, it is noteworthy that the posterior body 571 parts of mice are least vibrated among all body parts during lateral WB vibration (Fig. 3B 572 blue symbols, and 3C 2). This phenomenon indicates that the anterior and central body 573 parts remain relatively mobile, w h i l e t h e p o s t e r i o r p a r t s a r e comparatively stable and 574 immobile along the lateral axis. Such differential stability ma y play a critical role in 575 efficient control of body direction during locomotion. In mice, the hindlimbs are mainly 576 used to generate propulsion and accelerate the body, whereas the forelimbs can contribute 577 to braking and steering 41,42. Thus, posterior stability may facilitate propulsion, whereas 578 anterior mobility may be advantageous for steering. The mouse b ody may be analogous 579 to a rear-drive four-wheeled vehicle, complementing previously reported left-right 580 locomotor asymmetries 43. Taken together, axis-specifi c modulation of body vibration 581 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 23 control may represent an adaptive strategy that balances locomotor efficiency with 582 vibration attenuation to prevent body dysfunction. 583 584 Vibration discomfort in mice: relationships with body transmissibility and vibration 585 axis 586 Our behavioral results using the shuttle-box paradigm include t w o t y p e s o f 587 behaviors: escape, where mice move away from immediate and present aversive stimulus 588 (i.e., “Get me out of this vibration NOW”), and passive avoidan ce, where mice show 589 preventive behavior against predicted future threats (i.e., “Do not let me get into that 590 vibration”). Our analysis (Fig. 4; percentage of time spent in the vibration area) might 591 primarily quantify the passive avoidance component. Notably, ou r behavioral results 592 showed no carry-over effect, and the aversive vibration stimuli were not associated with 593 specific chamber locations (i.e., vibration areas) in mice (see Methods). Therefore, our 594 behavioral results likely repres ent vibration di scomfort corresponding to a mild, rather 595 than strong, unpleasant emotional state in mice. 596 In humans, it has been suggested that vibration discomfort can be related to body 597 transmissibility and resonan ce phenomena within the body 3,6. However, only weak 598 correlations have been found between vertical head motion and d iscomfort, and 599 psychophysically measured equivalent comfort contours do not mirror the inverse of body 600 transmissibility curves in humans 3. Rather, vibration frequency is considered the primary 601 determinant of human vibration discomfort 44. This is consistent with our findings that 602 vibration discomfort in mice depends on vertical vibration frequency when displacement 603 amplitude is fixed, irrespective of resonance frequencies at sp ecific body parts (Table 2; 604 Fig. 4C). Pain after WB vibration exposure also has been report ed to be frequency 605 dependent and independent of the resonance frequency in rats 45. Vibration discomfort is 606 thus a complex phenomenon determined by multiple internal and e xternal factors, and 607 cannot be predicted solely by the response of a single body par t or a single vibration 608 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 24 variable 3,44. Consequently, general models o f vibration discomfort remain e lusive, and 609 existing models have been developed under specific experimental conditions with distinct 610 constraints in humans 3. Notably, when vertical vibrati on acceleration is held constan t, 611 discomfort in mice becomes dependent on displacement amplitude rather than vibration 612 frequency (Fig, 4D), highlighting the dynamic interplay between frequency, displacement 613 amplitude, acceleration, and environmental context of vibration. 614 The present study demonstrated that vibration discomfort in mic e is specific to 615 vertical WB vibration (Fig. 4). Similar vertical axis specific phenomena have been 616 reported in humans, where vertical vibration induced greater di scomfort than horizontal 617 vibration (lateral or longitudi nal vibration) at frequencies of 4 Hz and above 44,46. This 618 indicates that heightened sensitivity to vertical vibration may be evolutionarily preserved 619 in mammals. Such axial (directio nal) dependency is reminiscent of a human cognitive 620 function, the inversion effect i n visual perception, such as Thatcher illusion (face 621 inversion effect: inverted faces r educe sensitivity to notice f ace distortion) 47 and the 622 shape-from-shading inversion effect (convex-concave perception changes, depending on 623 lighting cues based on the human visual system assumption that light comes from above) 624 48. Comparable face inversion effec ts along the vertical axis hav e been reported in non-625 human primates 49 and medaka fish 50. Because animals on earth are affected by gravity 626 and gravity always acts along the vertical axis, animals may have adapted themselves to 627 this constraint through evolution. Therefore, vertical WB vibration may possess a unique 628 biological significance in eliciting behaviors such as vibration avoidance in mice. 629 In conclusion, we identified axis-specific features of body pos ture control and 630 avoidance behavior under three-axis WB vibration in mice. It remains to be clarified how 631 such axial specificity is achiev ed from biomechanical and neuro physiological 632 perspectives. Interestingly, some parallels were observed betwe en vibration control in 633 mice and control strategies in four-wheeled vehicles. Analogies between quadrupeds and 634 vehicles 43 may provide a useful framework for future studies on the biome chanics and 635 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 25 the NMS system-mediated control of posture in mice. 636 637 638

639 640 Animals 641 All procedures for the care and treatment of animals were carried out in accordance with 642 the Japanese Act on the Welfare and Management of Animals and the Guidelines for the 643 Proper Conduct of Animal Experiments issued by the Science Coun cil of Japan. The 644 experimental protocols were approved by the Gunma University An imal Care and 645 Experimentation Committee (a pproval numbers: 18-019 and 23-018) . Adult C57BL/6 646 mice of both sexes (BW 17.3 – 25.3 g) were used for body part movement analysis, and 647 male mice (BW 21.8 – 31.7 g) were used for vibration avoidance (preference) tests. The 648 mice were maintained on a 12:12-hour light-dark cycle (lights o n a t 8 a m ) a n d a t a 649 constant temperature of 23±3℃. Rodent laboratory chow (MF, Oriental Yeast Co., Ltd) 650 and water were freely available in their home cages. 651 652 Vibration apparatus and video recording of body-part movements in awake and 653 anesthetized mice 654 Mice were exposed to sinusoidal base oscillations at frequencies ranging from 3 to 40 Hz 655 with a constant pe ak-to -peak displacement amplitude of 1 mm in three different axial 656 directions: vertical (up-down: z-axis), longitudinal (fore-aft: y-axis), lateral (left-right: x-657 axis). We utilized video object tracking to examine the vibration response in various parts 658 of the mouse body (see Methods below; Supplementary Videos 1 and 2). Unless otherwise 659 noted, the displacement amplitude of the vibration stimuli was set to 1 mm, as our 660 preliminary experiments showed that 1-mm vertical vibrations at 3-56 Hz for 1 hour had 661 no effect on the behavioral res ults of open field tests conduct ed immediately after 662 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 26 exposure to the vibrations (n = 4-10), suggesting that 1-mm vibrations are not harmful to 663 mice. Moreover, with 1-mm vibration stimuli, our video-based tracking method provided 664 sufficient spatial resolution to detect mouse body part movemen ts across a frequency 665 range of at least 3-25 Hz, which covered the resonance frequencies for all three vibration 666 axes in both awake and anesthetized conditions (Figs. 1-3; Supp lementary Figs. S1 and 667 S2). An electromagnetic vibratory actuator (m060, IMV CORPORATION, Osaka, Japan) 668 was used to generate vibration. The actuator was controlled by a vibration controller 669 system (K2, IMV Corporation, Osaka, Japan) connected with a sig nal amplifier (MA1, 670 IMV CORPORATION, Osaka, Japan). The magnitude (acceleration) an d frequency of 671 the base vibrations were monitored with an accelerometer (356A32, PCB Piezotronics, 672 NY , USA) placed on the vibrator platform. In experiments using awake mice, no sedatives 673 or anesthesia were administered. Mice were placed in a transparent acrylic chamber 674 (mouse box; 100 mm length × 40 mm width × 40 mm height) with a detachable lid 675 composed of a thin plastic sheet and thick polyurethane foam co ntaining multiple 676 ventilation holes. This chamber size allowed mice to maintain a natural hunched posture 677 continuously without stress. For vertical vibration, the chambe r was secured to the 678 actuator’s vibrating table with double-sided adhesive tape or w ith instant glue applied 679 onto polyimide tape, which was attached to both the table surface and the chamber bottom 680 (Fig. 1A). For longitudinal or lateral vibration, the actuator was rotated 90 degrees (i.e. 681 positioned horizontally), and an L-shaped attachment was screwe d onto the vibrator 682 platform. Depending on whether the rear or side panel of the mo use chamber was fixed 683 to the attachment, this setup enabled longitudinal or lateral vibration, respectively (Figs. 684 2A and 3A). A sheet of paper-type bedding (Pulmas 3000, Scitex, Kawasaki, Japan) was 685 placed on the chamber floor to absorb urine and feces. In exper iments under anesthesia, 686 mice were pretreated with a ketamine (100 mg/kg, body weight) and xylazine (10 mg/kg) 687 mixture dissolved in 0.9% NaCl s olution, administered intraperitoneally (100 µL/10 g). 688 To approximate a natural hunched posture, anesthetized mice were positioned with their 689 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 27 limbs folded under the body 9, and all paws were affixed directly to the vibrating table 690 (for vertical vibration) or to an attached acrylic plate (for l ongitudinal and lateral 691 vibration) using four strips of double-sided adhesive tape (Sup plementary Fig. S5B and 692 C). The positions of the adhesive stripes were fixed and bilate rally symmetric 693 (Supplementary Fig. S5A). Unless otherwise noted, mouse body mo vements were 694 recorded using the slow-motion video acquisition mode (1080p Full HD resolution at 240 695 frames per second [fps]) of an iP hone 8 Plus (Apple Inc., Cuper tino, CA, USA), 696 positioned laterally (for vertical and longitudinal vibrations; Figs. 1A and 2A, 697 Supplementary Fig. S5B, Supplementary Videos 1 and 2) or dorsally (for lateral vibration; 698 Fig. 3A, Supplementary Fig. S5C) relative to the mouse or the m ouse chamber. The 699 recording duration at each frequency was typically 1 min or les s (up to a maximum of 5 700 min), with inter-trial intervals of at least 0.5 min. The order of stimulus frequencies was 701 randomized. 702 When a frequency-modulated, single-sweep vibration stimulus was used, the 703 waveform was generated using a built-in function of the vibration controller system (K2, 704 IMV Corporation, Osaka, Japan). The stimulus frequency was linearly increased from 10 705 Hz to 30 Hz at a slow rate of 0.5 Hz/s, with a constant peak-to -peak displacement 706 amplitude of 1 mm (Supplementary Fig. S7A). 707 708 Measurements of displacement and phase differences at various body parts 709 To track body movements reliably, mouse body hair was shaved an d tracking points on 710 several body parts were marked directly on the skin using a whi teout pen prior to video 711 recording (Supplementary Fig. S5B and S5C; Supplementary Videos 1 and 2). In vertical 712 and longitudinal (fore-aft) vibration, measurements were performed at seven points along 713 the lateral side of the body: ‘head’, ‘shoulder’, ‘middle’, ‘back’, ‘lower back’, ‘tail base’, 714 and ‘heel’ (Figs. 1A and 2A). In lateral vibration, body-part m ovements were measured 715 from the dorsal side at seven regions arranged in rostrocaudal order: ‘forehead’, ‘head’, 716 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 28 ‘back1’, ‘middle’, ‘back2’, ‘lower back’, and ‘tail base’ (Fig. 3A). As reference markers 717 for vibration, solid objects suc h as the ‘mouse box’ affixed to the vibrating floor, were 718 marked with black dots using a black oil-based pen (Figs. 1A-3A ; Supplementary Fig. 719 S5B and S5C). 720 To focus on the steady-state response to vibration, body part m ovements were 721 analyzed at least 5 seconds after vibration onset. Because awake mice were able to move 722 freely in the chamber even during vibration, analysis was limit ed to video segments in 723 which the mouse maintained a stable hunched posture. Tracking o f the marked points in 724 the recorded videos was performed using the open-source softwar e Kinovea (version 725 0.8.26; https://www.kinovea.org), with the marker tool positioned at the center of each 726 marked point (Supplementary Videos 1 and 2). Of the measured coordinates (x and y in 727 pixels) over time (ms), the coordi nate axis either parallel or orthogonal to the vibration 728 direction was selected for analysis, depending on the direction of the vibration stimulus. 729 For example, in vertical vibrati on experiments, y-coordinate va lues were analyzed as a 730 function of time (Fig. 1B and 1D), whereas x-coordinate values were analyzed in Fig. 1E. 731 The values in pixels were converted to millimeters based on cal ibration scales derived 732 from known dimensions of objects such as the ‘mouse box’ and a yellow reference object 733 (Figs. 1A-3A; Supplementary Fig. S5B and S5C). Tracked data exh ibiting an unstable 734 baseline drift due to postural changes were excluded from the analysis. 735 From the tracked data segments containing at least stable six cycles of vibration 736 response at each body part, displacement peak-to-peak amplitudes (i.e., double 737 amplitudes) and phase differences at each body part relative to a reference, were analyzed 738 using a custom-made macro of Igor Pro 8.04 software (WaveMetrics, Lake Oswego, OR, 739 USA) by NH (Supplementary Fig. S1). Positive and negative peak positions were 740 detected using the built-in “FindPeak” function in Igor Pro (box size for sliding average: 741 4-6), and all detections were visually confirmed (Supplementary Fig. S1, cross symbols). 742 Peak-to-peak displacement was calculated as the difference between adjacent positive and 743 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 29 negative peaks of the displacemen t waveform (Supplementary Fig. S1, black double-744 headed arrows). All displacement amplitudes were normalized to that of the ‘reference’ 745 in each trial. Phase difference was defined as the delay relative to the adjacent ‘reference’ 746 peak. Specifically, the delay was initially calculated as the t ime difference between the 747 displacement peak (positive or n egative) of the ‘reference’ (Supplementary Fig. S1; red 748 vertical dotted lines) and the nearest subsequent peak of the s ame polarity in the 749 displacement waveform of each body part (Supplementary Fig. S1; black vertical dotted 750 lines). This delay (Supplementary Fig. S1; red arrows) was then converted to an angle 751 (degrees) based on the cycle time of the vibration frequency (with a full cycle time delay 752 corresponding to -360 degrees). For each mouse, averaged values of displacement and 753 phase difference were calculated from at least six successive c ycles of the displacement 754 waveform. At higher vibration frequencies, where displacement amplitudes were often 755 greatly reduced (i.e., strongly damped) and approached the detection limit, the waveforms 756 became nearly flat, making peak detection unreliable. This led to large variability in the 757 measured phase differences and compromised the accuracy of vibr ation-response 758 quantification. Such unreliable da ta were excluded from further analysis, resulting in 759 smaller sample sizes at higher frequencies. All data are presented as mean ± SEM (awake 760 mice, n = 2-10; anesthetized mice, n = 4-10). Due to technical limitations of the vibrator 761 at low frequencies, vibrations at around 3 Hz could not be precisely controlled. Although 762 the vibrator was set to generate vibrations with 1-mm amplitude at 3 Hz, the actual 763 ‘reference’ displacement was reduced slightly (Supplementary Fig. S6). 764 To examine the frequency-dependent progressive increment of phase difference 765 and the interrelationship between left and right body parts, we utilized the single-sweep 766 frequency-modulated lateral WB vi bration stimulus (see Methods above). Symmetrical 767 left and right posterior body parts were designated as measurement points, and body part 768 movements were recorded from the rear view (Supplementary Fig. S8A). Measurements 769 of stimulus frequencies, displace ments and phase delays were de scribed in 770 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 30 Supplementary Figure S7 and perfo rmed using a custom script of MATLAB (version 771 9.9.0, R2020b; The MathWorks Inc., Natick, MA, USA) written by M. Suzuki. The results 772 obtained with the frequency-modulated stimulus were largely com parable to those 773 obtained with the fixed frequenc y stimulus (Supplementary Fig. S8), indicating that the 774 modulation rate was sufficiently slow to allow steady-state responses of the body parts. 775 776 Model fitting 777 To describe the dynamic response of the mouse body, a simple single-degree-of-freedom 778 (SDOF) vibration model with lumped parameters was employed. This model consists of 779 a single moving mass supported by a spring and damper, excited by base displacement 780 (Supplementary Fig. S4) 3,4. For model fitting, we focused on the displacement response 781 of the ‘middle’ body part in aw ake mice under vertical vibratio n, as a simple and 782 reasonable approximation of the whole-body dynamics. The ‘middl e’ body part was 783 considered to approximate the center of the total body mass ( m, in kg). The mass 784 parameter m was fixed at 0.022 kg (22.0 g), corresponding to the typical b ody weight of 785 an adult mouse. In the passive control model of base excitation (i.e., vibration applied at 786 the base), both the spring coefficient ( k, in N/m) and damping coefficient ( c, in N·s/m) 787 were assumed to be constant. According to the model, the relative amplitude of the mass 788 displacement with respect to the base (input) displacement (i.e., transmissibility) is given 789 as a function of the stimulus frequency (f, in Hz) as follows: 790 Relative amplitude ൌ ඩ ଵାቀଶ∙௓𝒇 ೑೙ ቁ మ ൜ଵିቀ𝒇 ೑೙ ቁ మ ൠ మ ାቀଶ∙௓𝒇 ೑೙ ቁ మ [ 1 ] 791 where fn is the natural frequency, and Z is the damping ratio, defined as: 792 𝑓௡ ൌ ଵ ଶగට𝒌 𝒎 ,𝑍 ൌ 𝒄 ଶ√𝒎𝒌. [ 2 ] 793 Model fitting was performed to the experimental data using this equation, treating k and 794 c as free parameters (Supplementa ry Fig. S4A). In the second mod el, an active control 795 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 31 approach incorporating a dynamically controllable damper was em ployed. Specifically, 796 we adopted an ideal skyhook damping control system (Supplementary Fig. S4B) in which 797 the damping force is regulated via feedback based on the absolute velocity of the mass 34. 798 This “skyhook” scheme was chosen for its simplicity in mathematical formulation. In this 799 model, the damping coefficient csky serves as a feedback gain proportional to the absolute 800 velocity. In this model, the relative amplitude is expressed as: 801 Relative amplitude ൌ ଵ ඨ൜ଵିቀ𝒇 ೑೙ ቁ మ ൠ మ ାቀଶ∙௓ೞೖ೤ 𝒇 ೑೙ ቁ మ . [ 3 ] 802 Here, Zsky is defined in the same way as Z, except that the damping coefficient, c i s 803 replaced with csky. The model was fitted to th e experimental data using k and csky as free 804 parameters (Supplementary Fig. S4B). Model fitting was limited to data from awake mice, 805 as data from anesthetized mice exhibited complex responses suggestive of a multi-degree-806 of-freedom system at higher fre quencies (Supplementary Fig. S3) , which made them 807 unsuitable for the simplified models described above. In additi on, model fitting was not 808 performed to data from lateral or longitudinal WB vibrations, because these conditions 809 were also complex, indicative of a multi-degree-of-freedom syst em, and likely violated 810 the assumption of a simple SDOF system. 811 812 Preference (avoidance) test for vibration 813 All animals were habituated to experimenters by daily handling for 5 days prior to 814 behavioral testing. During handling, each mouse was placed on t he gloved palm of the 815 experimenters for 10 minutes per day. 816 To examine vibration preference ( o r a v o i d a n c e ) i n m i c e , a t w o - chambered 817 shuttle-box behavioral paradigm was employed (Fig. 4A). The apparatus consisted of two 818 open-ended acrylic rectangular cha mbers of identical size (130 mm [length] × 45 mm 819 [width] × 45 mm [height]). The top panels were transparent to a llow overhead video 820 recording of the mouse, while the remaining panels were opaque grey to reduce exposure 821 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 32 to external visual stimuli. The c hambers were arranged such tha t their open ends faced 822 each other, separated by a 2-mm gap, allowing the mouse to shuttle freely between them. 823 One chamber was affixed to the vibrating table (vibrator area), while the other 824 was placed on an immobile platform of identical height (static area). A sheet-type white 825 bedding (Pulmas3000, Scitex, Kawasaki, Japan) was taped to the floors of both chambers 826 to absorb urine and feces, and to enhance color-based detection of the mouse body (black-827 colored) during image analysis. Illumination intensity ranged from 245 to 432 lux, and 828 the brightness levels in both chambers was adjusted to the simi lar level so that the 829 difference between them was less than 10 lux. All tests were co nducted in a quiet room 830 between 9 a.m. and 12 p.m. For each mouse, test sessions were s eparated by at least one 831 day. 832 Before testing, mice were transported to the testing room and acclimated for 20 833 minutes in a plastic cage (260 mm [length] × 155 mm [width] × 125 mm [height]) housing 834 three mice per cage. Following acclimation, each mouse was plac ed in the vibration 835 chamber (Fig. 4A; vibrator area), where it voluntarily entered without signs of distress. 836 The opening between the two chambers was closed with a removabl e guillotine-style 837 partition, and the vibration stimulus was applied for 1 minute to allow sufficient exposure 838 for preference formation. Subseque ntly, the partition was remov ed, and the mouse was 839 allowed to move freely between the two chambers for 30 minutes. Each mouse was tested 840 once per day with an interval of at least one day between trial s. On the first day, a 841 habituation trial was conducted without vibration to allow mice to become familiar with 842 the shuttle-box environment. From day 2 onward, mice were exposed to vibration stimuli 843 presented in randomized order. The vibration conditions include d a fixed peak-to-peak 844 displacement amplitude of 1 mm at different frequencies (6, 12, 16, 32, and 40 Hz), as 845 well as a no-vibration control, a nd were tested separately in t hree different directions 846 (vertical, longitudinal [fore–aft], and lateral). Among these conditions, peak acceleration 847 varied depending on frequency. Un like active avoidance tests wi th electrical footshock 848 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 33 (fear conditioning), in which the chamber is associated with th e aversive stimulus, our 849 test protocol produced no carryover effect (i.e., the avoidance result in a previous trial did 850 not affect behavior in the following trial). In a separate experiment using vertical vibration, 851 the peak-to-peak acceleration was fixed at 5.0532 m/s², and the displacement amplitude 852 was varied across different frequencies (5.1, 6, 16, and 32 Hz) with corresponding 853 amplitudes of 9.84, 7.11, 1.00, and 0.25 mm, respectively (Fig. 4D). Between trials, the 854 chambers were cleaned using 70% ethanol and immersed in a dilut ed chlorine bleach 855 solution for at least 30 minutes to eliminate residual olfactory cues. 856 Mouse behavior was recorded at 29.97 fps (640 × 480 pixels; VGA resolution) 857 using a digital camera (EX-ZR1000, CASIO, Tokyo, Japan) placed above the chambers 858 (Fig. 4A). For image analysis, the video frame rate was downsampled to ~5 fps (one frame 859 every 200 ms), and the analysis area was limited to the two cha mbers. A custom Python 860 (version 3.8) program written by NH was used to analyze mouse p osition. The mouse 861 body was detected as a binarized object (i.e., a chunk of pixels) using HSV color detection 862 (with black representing the mouse body), followed by morphological filtering (opening 863 and closing operations using a 4×4 kernel) implemented with Ope nCV (version 4.10). 864 The mouse position was defined as the centroid of the detected object in each frame (Fig. 865 4A, asterisks), and detection ac curacy was confirmed by visual inspection. The 866 percentage of time spent in the vibrator area was calculated us ing frame interval and the 867 number of frames in which the centroid position was located in either “vibrator area” or 868 “static area”. Mouse positional probabilities in the two chambe rs during the preference 869 test were visualized as heatmaps of bivariate kernel density estimation (KDE) plots using 870 a Python data visualization library, Seaborn (version 0.10.1). 871 872 Statistical analysis and use of a Large Language Model 873 Statistical significance was assessed with the Mann-Whitney U-t est for 874 comparisons between awake and anesthetized mice (Figs. 1-3 and Supplementary Fig. 875 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 34 S2; Holm-Sidak correction) and Dunn’s multiple comparison test for evaluating vibration 876 preference relative to each cor responding control (Fig. 4), usi ng GraphPad Prism 10 877 software (GraphPad, La Jolla, CA). Differences were considered statistically significant 878 at P < 0.05. Pooled data are presented as mean ± SEM. ChatGPT (OpenAI, USA) was 879 used to assist with grammatical editing and language refinement after the manuscript was 880 drafted. 881 882 883

884 We thank Yumi Fukuzaki for assistance with preliminary experim ents and 885 Sakura Shimakata for help with behavioral experiments. We are also grateful to Mamoru 886 Sawada for invaluable comments and discussions. 887 888 889 Author contributions 890 N.H. conceived the project, and M. Suzuki and N.H. designed and performed the 891 experiments. M. Suzuki and N.H. a nalyzed and interpreted the da ta with help from M. 892 Saito, T.U., H.I., K.S., and H.H. M. Suzuki and N.H. drafted th e manuscript, and N.H. 893 completed the paper with inputs from all the authors. 894 895 896 Competing interests 897 This study was supported by S UBARU CORPORATION. 898 899 900

& Correspondence 901 Correspondence and requests for materials should be addressed to Hirokazu 902 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 35 Hirai or Nobutake Hosoi. 903 904 905 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 36 Tables 906 907 Table 1. Rough estimation of stiffness (spring constants), calculated based on a 908 single-degree-of-freedom system excited by vertical vibration, using fixed body part 909 weights and their corresponding resonance frequencies in anesth etized and awake 910 mice. 911 Rough estimation of stiffness, k (N/m) middle head Anesthetized 347.4 44.6 Awake 146.8 16.1 Stiffness k (i.e., the spring constant) was estimated using the equation: k = m • (2π•fn)2, 912 where m is the mass, and fn is the natural frequency of each body part in response to 913 vertical vibration. The measured resonance frequency (Fig. 1C1) was assumed to roughly 914 approximate the natural frequency (fn) of the system. However, this assumption may lead 915 to an underestimation of k, as resonance frequencies in damped systems are typically 916 slightly lower than their corresponding natural frequencies. Th e body part ‘middle’ was 917 approximated as the center of mass, and its mass m was fixed at 22.0 g (0.022 kg), 918 representing the typical body weight of an adult mouse. The mass of the ‘head’ was fixed 919 at 2.8 g (0.0028 kg), based on the mean ratio of head weight to total body weight in mice 920 (0.128 ± 0.005, n = 5) and the body weight value described above. 921 922 923 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 37 924 Table 2. Frequency- and vibration axis-dependent displacement of the ‘middle’ and 925 ‘head’ body parts in awake mice in response to 1-mm vibrations. 926 Displacement at ‘middle’ Displacement at ‘head’ Vibration frequency (Hz) Vibration frequency (Hz) Vibratory excitation direction 6 12 16 32 40 6 12 16 32 40 Vertical (up-down) – ↑↑ ↑↑ ↓↓ ↓↓ – ↑ ↓ ↓↓ ↓↓ Longitudinal (fore-aft) ↑ – ↓↓ ↓↓ ↓↓ ↑ – ↓ ↓↓ ↓↓ Lateral (left-right) ↑↑ ↑↑ – ↓↓ ↓↓ ↑↑ – ↓ ↓↓ ↓↓ Abbreviations ‘–‘, ‘↑’, ‘↑↑’, ‘↓’, and ‘↓↓’ indicate the relative magnitude of displacement 927 in response to vibration along its excitation axis. The ‘middle’ body part is approximated 928 as the center of mass of the mous e. Specifically, ‘–‘, ‘↑’, ‘↑↑ ’, ‘↓’, and ‘↓↓’ represent 929 displacements that are equal to, slightly greater than, much greater than, slightly smaller 930 than, and much smaller than (i.e., strongly damped) the stimulu s amplitude (1 mm), 931 respectively. 932 933 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 38 Figures 934 935 Figure 1. Differential movements of mouse body parts in respons e to vertical WB 936 vibrations in awake and anesthetized mice. 937 (A) Illustration of the experimental setup and the body part lo cations (tracking points) 938 used to measure displacements during vertical WB vibration (sin usoidal waveform, 1-939 mm peak-to-peak amplitude at vari able frequencies). Color-coded symbols indicate the 940 corresponding mouse body parts analyzed in the other panels (B, D and E). A mouse was 941 placed in a small transparent rectangular acrylic chamber (‘mou se box’) affixed to a 942 vibrator. Body movements were r ecorded from the side with a dig ital camera at 1080p 943 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 39 (Full HD) resolution and 240 fps during vibration exposure (see Supplementary Video 1). 944 (B) Relative displacement amplit udes of each body part, as depi cted in (A), during 945 vertical WB vibration. Measured displacement at each body part was normalized to that 946 of the reference (mouse box), a nd the resulting relative amplit udes were averaged in 947 awake (left) and anesthetized (right) mice (see Methods). 948 (C1 and C 2) Resonance frequency (C 1; defined as the frequency at which the relative 949 displacement amplitude was maximum in the averaged frequency re sponse profiles 950 shown in (B)), and corresponding maximum relative displacement amplitudes (C2) of 951 each body part in awake (filled symbols; n = 6-9) and anesthetized (open symbols; n = 4-952 10) mice (Mann-Whitney tests between awake and anesthetized mic e, **P < 0.01, *P 953 <0.05; see also Supplementary Fig. S2). Resonance frequencies w ere lower in all body 954 parts of awake mice compared to anesthetized mice. Notably, the m a x i m u m r e l a t i v e 955 displacement amplitudes of anterior body parts (head, shoulder, and middle) were reduced 956 in awake mice, whereas those of posterior parts (lower back, ta il base, and heel) were 957 comparable between the two groups. 958 (D) Mean phase difference of p eriodic displacement in each body part relative to the 959

point in awake (left) and anesthetized (right) mice. Phase difference was first 960 measured as the time lag between t he displacement peak of the r eference and the 961 temporally adjacent displacement peak of each body part. This t ime lag was then 962 converted to a phase angle (in degrees), with negative values i ndicating a phase delay, 963 based on the cycle duration of the vibration frequency (see Methods). 964 (E) Mean relative displacement amplitudes of each body part mea sured along the 965 longitudinal (fore-aft) axis, which is orthogonal to the vertical vibration axis (i.e., 966 stimulus axis), in awake mice. Displacements along the fore-aft axis were normalized to 967 the vertical displacement amplitude of the reference point (i.e., stimulus input amplitude). 968 Error bars represent the standard error of the mean (SEM) in this and subsequent figures. 969 In (B), (C2), and (E), horizontal broken lines indicate the reference amplitude level (equal 970 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 40 to the vibration stimulus). In (D), broken lines at -90 and -180 degrees indicate a quarter-971 cycle and a half-cycle phase delay, respectively. 972 973 974 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 41 975 Figure 2. Minor difference in the movements of mouse body parts in r esponse to 976 longitudinal (i.e., fore-aft) WB vibrations in awake and anesthetized mice. 977 (A) The experimental setup and measurement locations (tracking points) for mouse body 978 movements were similar to those of Figure 1, except that the vi brator was tilted 979 horizontally. The rear plane of the mouse box was attached to t he vibrator to apply 980 longitudinal WB vibration (sinusoidal, 1 mm peak-to-peak amplit ude with variable 981 frequency) to mice. 982 (B-E) Data are presented in the same way as in Figure 1. Notabl y, the reduction in 983 resonance frequency at the body parts of awake mice was minor, especially in the medial 984 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 42 parts of the body (the ‘middle’ and ‘back’) and the ‘heel’ (C1). The difference in maximum 985 relative displacement between awake (n = 5-6) and anesthetized (n = 6-10) mice was also 986 minor across body parts, except fo r the ‘lower back’, and ‘tail b a s e ’ ( C2). During 987 longitudinal WB vibration in awake mice, pitching of the mouse body was observed, with 988 pronounced movements along the orthogonal vertical axis in the anterior and posterior 989 parts of the body (E). 990 991 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 43 992 993 Figure 3. Large difference in the movements of mouse body parts in response to 994 lateral WB vibrations between awake and anesthetized mice. 995 (A) The experimental setup was si milar to that shown in Figure 1, except that the side 996 plane of the mouse box was affixed to the horizontally tilted v ibrator, allowing the 997 application of lateral WB vibration (sinusoidal, 1 mm peak-to-p eak amplitude with 998 variable frequency) to mice. Moreover, tracking points for mouse body movements were 999 placed along the midline of the body, and movements were recorded from above. 1000 (B-E) Data are presented in the same way as in Figure 1. As with vertical WB vibrations 1001 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 44 (Fig. 1), resonance frequencies in most body parts were reduced in awake mice compared 1002 to anesthetized mice, except for the ‘tail base’ (C1). The difference in maximum relative 1003 displacement between awake and anesthetized mice was markedly l arger than that 1004 observed in the other axial WB vibrations (C 2; Mann-Whitney U-test, ** P < 0.01, 1005 comparison between awake [n = 6-7] and anesthetized [n = 6] groups). Movements of the 1006 body parts along the orthogonal longitudinal (i.e., fore-aft) a xis were negligible during 1007 lateral WB vibrations (E). 1008 1009 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 45 1010 Figure 4. Mice selectively avoid vertical vibrations, but not l ongitudinal or lateral 1011 ones, depending on frequency and amplitude. 1012 (A) Two mouse chambers were placed with their open ends facing each other and 1013 separated by a 2-mm gap. One chamber was attached to the vibrator (upper panel, vibrator 1014 area), while the other was isola ted from vibration (upper panel , static area). Mouse 1015 positions were recorded with a digital video camera for 30 minu tes to assess approach-1016 avoidance behavior in response to various vibration conditions. The upper panel shows a 1017 cropped representative video frame, and the lower panel shows the estimated mouse 1018 position (indicated by a red asterisk), determined as the centroid of the mouse body (i.e. 1019 a chunk of white pixels) using a color-detection image processing method (see Methods). 1020 Broken lines indicate the boundaries between the static and the vibrator areas. 1021 (B) Heatmaps showing the mouse’s positional probability for 30 minutes in response to 1022 three-axis vibrations. Warmer c olors indicate higher probabilit y of presence at each 1023 location. Broken lines represent the boundaries between the static and the vibrator areas. 1024 White numbers indicate the percentage of time spent in the vibrator area. 1025 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 46 (C) Mean percentages of time spent in the vibrator area over a 30-min test period, plotted 1026 against vibration frequency (1- mm peak-to-peak amplitude) for t hree vibration axes: 1027 vertical (n = 18-21 mice), longitudinal and lateral (n = 18 eac h). Control data were 1028 obtained under no-vibration cond itions (vertical, n = 21; longi tudinal and lateral, n = 9 1029 each). Dunn’s multiple comparison test, ** P < 0.0001, ‡P < 0.005 compared with the 1030 corresponding control for each vibration condition. Mice specif ically avoid vertical 1031 vibrations at frequencies above 30 Hz. 1032 (D) Mice also avoided vertical vibrations of larger amplitude at frequencies below 16 Hz, 1033 even under a constant peak acceleration of 5.0532 m/s2, equivalent to that generated by a 1034 vibration stimulus with a 1.00 mm peak-to-peak displacement at 16 Hz (n = 15 mice in 1035 each condition; Dunn’s multiple comparison test, **P < 0.0001, ‡P < 0.005 compared to 1036 16 Hz, 1.00 mm or 32 Hz, 0.25 mm). The dotted lines in (C) and (D) indicate chance level. 1037 1038 1039 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 47 Supplementary Figures 1040 1041 1042 Supplementary Figure S1. Method for measuring peak-to-peak amplitude and time 1043 delay (phase difference) from tracked displacement of mouse body parts. 1044 Representative displacement waveforms of the ‘lower back’, ‘sho ulder’, and ‘reference 1045 (mouse box)’ during vertical WB vibration (sinusoidal, 15 Hz, 1 mm amplitude) in awake 1046 mice are shown to illustrate how amplitude and time delay (phas e difference) were 1047 measured from the displacement waveforms. Notably, compared to the ‘reference (mouse 1048 box)’, the displacement of the ‘lower back’ was larger (i.e., resonance), while that of the 1049 ‘shoulder’ was slightly smaller un der this exper imental condition. For measurement, a 1050 stable period of the displacement waveforms was selected so that the waveforms 1051 contained at least six consecutive vibration cycles without pos tural change. Cross 1052 symbols indicate the detected positive and negative peak locati ons, which were used to 1053 measure peak-to-peak amplitude (double amplitude; black double- headed arrows) and 1054 time delay (red arrows) (see Methods). Within the stable period , multiple measurements 1055 of peak-to-peak amplitude were made, and multiple time delays w ere obtained based on 1056 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 48 time points of the reference peaks (vertical red dotted lines) and the corresponding peaks 1057 of each part part (vertical black dotted lines). These values w ere averaged to represent 1058 individual data points. Averaged time delays were converted to angular values based on 1059 the cycle time of the vibration frequency for further analysis. 1060 1061 1062 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 49 1063 Supplementary Figure S2. Resonance frequencies and maximum rela tive 1064 displacement amplitudes derived from individual mice. (A-C) In contrast to the group-1065 averaged profiles shown in Figs. 1-3, resonance frequencies and maximum relative 1066 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 50 displacement amplitudes were obtained from individual frequency–amplitude profiles at 1067 each body part of individual mice. These values were then avera ged across animals, and 1068 the mean values were plotted against each body part (left panel , resonance frequency; 1069 right panel, maximum relative displacement amplitude). (A), (B) , and (C) correspond to 1070 vertical, longitudinal, and lateral WB vibrations, respectively. ***P < 0.005, **P < 0.01, 1071 and *P < 0.05 (multiple Mann-Whitney tests between awake and anesthetized mice). 1072 1073 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 51 Supplementary Figure S3. Displacement time courses at various body parts in 1074 anesthetized mice during vertical WB vibration at 40 Hz. (A-C) Displacement along 1075 the vertical vibration axis at each body part (Fig. 1A) was monitored over time in 1076 anesthetized mice exposed to vertical WB vibration at 40 Hz (sinusoidal, 1-mm peak-to-1077 peak amplitude) using video tracking (see Methods). Panels (A), (B), and (C) show data 1078 from three different mice. The y-axis scale of the leftmost ‘reference’ graph corresponds 1079 to the scales used for the other body parts within each panel. In contrast to the 1080 ‘reference’, the displacement waveforms in all body parts except for the ‘heel’, 1081 exhibited complex patterns with pronounced positive-negative asymmetry and large 1082 inter-individual variability, making it difficult to reliably measure displacement 1083 amplitude and phase difference (see Methods). In one typical pattern of the complex 1084 displacement waveform, a relatively small positive-going hump appeared during the 1085 downward phase of displacement, as seen at the ‘tail base’ and ‘shoulder’ in (A) and 1086 (B). In another typical pattern seen at the ‘tail base’, ‘lower back’, and ‘middle’ in (C), 1087 two positive peaks appeared to merge, resulting in a broader and larger displacement 1088 waveform with a doubled apparent cycle time (~50 ms) compared to the 25-ms cycle 1089 time of the ‘reference’ trace. 1090 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 52 1091 1092 Supplementary Figure S4. An active control model (skyhook dampe r control) fits 1093 the displacement of the mouse body part ‘middle’ in response to vertical vibration 1094 better than a passive control model. (A) and (B) show the two vibration control models, 1095 passive control and active control (upper panels), and their co rresponding model-based 1096 fits to the same experimental data (lower panels), respectively . The measured 1097 displacement of the ‘middle’ body part in response to vertical whole-body vibration (Fig. 1098 1) was normalized to the reference amplitude at each frequency (i.e., amplitude of the 1099 basal vibration stimulus) and us ed for model fitting (green fil led circles), because the 1100 ‘middle’ is approximated as the center of the total lumped mouse body mass. Both models 1101 are based on a single-degree-of-freedom vibration system consis ting of a lumped mass 1102 (m) connected in parallel to a spring (stiffness, k) and a damper (damping coefficient, c). 1103 In the upper panels, x(t) represents the displacement of the mass at time t, and red double-1104 headed arrows indicate the basal vertical vibration input. The mass was fixed at 22.0 g 1105 (0.022 kg, corresponding to the typical body weight of an adult mouse), while k and c (or 1106 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 53 csky) were treated as free parameters for model fitting (see Methods). In the active control 1107 model (B), the damper was assumed to be feedback-controlled (i. e., a controllable 1108 damping coefficient, ccontrollable) and a simple skyhook control scheme was adopted (B, 1109 upper panel). In this model, the controlled damping coefficient, csky is regulated based on 1110 the absolute velocity of the mas s. The passive control model yi elded estimates of c = 1111 2.071 and k = 144.71 (A, lower panel, black line), while the skyhook contr ol model 1112 provided improved fitting with csky = 1.938 and k = 239.78 (B, lower panel, blue line). 1113 1114 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 54 1115 Supplementary Figure S5. Illustrations of paw positioning in an esthetized mice to 1116 approximate a hunched posture. (A) Layout of four double-sided adhesive strips 1117 (orange rectangles; centers indic ated by filled circles) used t o affix all paws of the 1118 anesthetized mouse. This setup enabled the limbs to be folded u nder the body and kept 1119 the mouse fixed to the vibrating floor. (B and C) Illustrations including rulers with 1-mm 1120 scales and examples of mice affi xed to the vibrating floor. In the vertical vibration 1121 condition (B), a yellow object with a black-marked dot at its b ase was taped to the floor 1122 as a reference marker for stimulation. In the lateral vibration condition (C), black-marked 1123 dots on the floor near the ‘face’ and ‘tail base’ were used as reference points for vibration. 1124 1125 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 55 1126 Supplementary Figure S6. Displacement amplitudes at the reference points in 1127 various vibration conditions with different frequencies. The vibrator was configured 1128 to generate vibration stimuli with a peak-to-peak amplitude of 1 mm. To verify the 1129 accuracy of the generated vibrat ion stimuli, actual displacemen t amplitudes were 1130 measured at reference points marked on the frame of the mouse c hamber (see Figs. 1A–1131 3A), on yellow reference objects affixed to the vibration platf orm (Supplementary Fig. 1132 S5B), or directly on the vibratio n platform itself (Supplementa ry Fig. S5C). At 1133 frequencies above 5 Hz, the measured amplitudes closely reached the intended 1-mm 1134 level (indicated by the dashed line), suggesting that the vibra tion stimuli were well 1135 controlled well in this range, regardless of vibration axis. In contrast, at lower frequencies 1136 (3–5 Hz), the displacement amplitudes fell below 1 mm due to te chnical limitations of 1137 the vibrator system. 1138 1139 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 56 1140 Supplementary Figure S7. Method for measuring peak-to-peak disp lacement 1141 amplitude and phase delay of posterior body parts using a slowly frequency-1142 modulated single-sweep lateral (l eft-right) vibration stimulus i n a w a k e m i c e . (A) 1143 Schematic illustration of two ideal displacement waveforms (reference and body part) in 1144 response to a frequency-modulated lateral vibration. For illustrative clarity, the frequency 1145 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 57 modulation is shown at a 14 times faster rate than the actual rate used in the experiments. 1146 Positive and negative peaks (cross symbols) were detected from all the waveforms. Time 1147 windows between two neighboring peaks with the same polarity we re defined as 1148 negative-peak windows (light blu e band) and positive-peak windo ws (cream band). For 1149 each time window, the vibration frequency was calculated from the reference trace as the 1150 inverse of the interval between successive peaks of the same po larity (Tₙ and Tₚ). Peak-1151 to-peak displacement amplitudes of each trace were measured wit hin each window, 1152 averaged, and associated with the corresponding frequency (left panel). For phase delay 1153 analysis (right panel), the time difference between the peak in the reference and the one 1154 in body-part trace (red arrowheads) was measured, converted to angular phase delay, and 1155 assigned to the measured frequency in each time window. Data we re grouped into 0.25 1156 Hz bins, and the mean was calculated for each bin. (B) Five representative displacement 1157 traces are shown: one reference trace (measured along the lateral axis) used to determine 1158 vibration frequency, and four body part traces (left/right ‘upp er hip’ and ‘back thigh’) 1159 measured along the vertical axis. Phase delay between left and right body parts was 1160 determined by measuring the time difference between correspondi ng peaks (the same 1161 polarity) of left and right traces in each time window (red and magenta arrows in cream 1162 and light blue bands, respectively), assigning the result to the calculated frequency. Data 1163 were binned and averaged as described in (A). 1164 1165 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 58 1166 Supplementary Figure S8. Frequenc y-modulated vibration stimulus reveals 1167 progressive development of phase differences exceeding -180° an d antiphase 1168 movements between left and right body parts during lateral WB vibration in awake 1169 mice. 1170 (A) The experimental setup was similar to that in Figure 3, except that the movements of 1171 posterior body parts were recorded from behind during a frequency-modulated lateral WB 1172 vibration. A single-sweep frequency-modulated lateral vibration stimulus (10–30 Hz at 1173 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 59 0.5 Hz/s, peak-to-peak displacement amplitude of 1 mm) was used to continuously assess 1174 frequency-dependent changes in phase difference. Symmetrical left and right body parts 1175 were labelled as measurement points to examine interlimb moveme nt relationships. 1176 Color-coded symbols and lines correspond to specific left or ri ght body parts and the 1177 associated analysis results in the other panels (B-E). 1178 (B-E) Solid and dotted lines indicate data obtained from the fr equency-modulated 1179 vibration stimulus (see Supplemen tary Figure S7), and discrete symbols represent data 1180 from frequency-fixed lateral vibration stimuli (see Supplementa ry Figure S1). The 1181 consistency between the two data sets supports the validity of the frequency-modulated 1182 stimulus analysis. Data are presented in the same way as in Fig ure 3, except that 1183 displacement amplitudes were not normalized to the ‘reference’ displacement. (C) Phase 1184 differences along the lateral vibration axis gradu ally increased with stimulus frequency 1185 in upper body parts (‘top’ and ‘upper hip’), eventually exceeding -270°, whereas in lower 1186 body parts (‘back thigh’ and ‘heel’), phase differences remaine d below -180°. (D) 1187 Displacement along the vertical axis (orthogonal to the lateral stimulus axis) was 1188 substantial, indicating a multip le-degree-of-freedom response system. (E) Phase 1189 difference analysis along the vertical axis between left and ri ght posterior body parts 1190 revealed a constant antiphase relationship (approximately -180° ) across vibration 1191 frequencies, suggesting that these left and right body parts move in a rotational, seesaw-1192 like manner during lateral WB vibration. 1193 1194 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 60 1195 Supplementary Video 1. Mouse body movements in response to vert ical WB 1196 vibration in awake mice. 1197 Vibration parameters: sinusoidal displacement, 1-mm peak-to-peak amplitude at 16 Hz. 1198 Playback speed is ~10 times slower than real time. Colored lines represent the movement 1199 trajectories of different body parts. 1200 1201 Supplementary Video 2. Mouse body movements in response to long itudinal (fore-1202 aft) WB vibration in awake mice. 1203 Vibration parameters: sinusoidal displacement, 1-mm peak-to-peak amplitude at 14 Hz. 1204 Playback speed is ~10 times slower than real time. Colored lines represent the movement 1205 trajectories of different body parts. 1206 1207 1208 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 61

1209 1210 1. Horak, F. B. Postural orientation and equilibrium: what do w e need to know about 1211 neural control of balance to prevent falls? Age Ageing 35 Suppl 2, ii7–ii11 (2006). 1212 2. Ivanenko, Y . & Gurfinkel, V . S. Human postural control. Front. Neurosci. 12, 171 1213 (2018). 1214 3. Griffin, M. J. Handbook of Human Vibration . (Academic Press, San Diego, CA, 1215 1990). 1216 4. Mansfield, N. J. Human Response to Vibration . (CRC Press, London, England, 1217 2005). 1218 5. Dupuis, H. & Zerlett, G. The Effects of Whole-Body Vibration . (Springer-Verlag 1219 Berlin Heidelberg, 1986). 1220 6. Taiar, R., Machado, C. B., Chiementin, X. & Bernardo-Filho, M. Whole Body 1221 Vibrations: Physical and Biologi cal Effects on the Human Body . (CRC Press, 1222 Philadelphia, PA, 2018). 1223 7. Reynolds, R., Garner, A. & Norton, J. Sound and Vibration as Research Variables in 1224 Terrestrial Vertebrate Models. ILAR J. 60, 159–174 (2020). 1225 8. Reynolds, R. P., Li, Y ., Garner, A. & Norton, J. N. Vibratio n in mice: A review of 1226 comparative effects and use in translational research. Animal Model Exp Med 1, 116–1227 124 (2018). 1228 9. Rabey, K. N., Li, Y ., Norton, J. N., Reynolds, R. P. & Schmitt, D. Vibrating 1229 Frequency Thresholds in Mice and Rats: Implications for the Eff ects of Vibrations 1230 on Animal Health. Ann. Biomed. Eng. 43, 1957–1964 (2015). 1231 10. Zeeman, M. E. et al. Whole-body vibration at thoracic resonance induces sustained 1232 pain and widespread cervical neuroinflammation in the rat. Clin. Orthop. Relat. Res. 1233 473, 2936–2947 (2015). 1234 11. Prisby, R. D., Lafage-Proust, M.-H., Malaval, L., Belli, A. & V ic o , L . E f fe c ts o f 1235 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 62 whole body vibration on the skele ton and other organ systems in man and animal 1236 models: what we know and what we need to know. Ageing Res. Rev. 7, 319–329 1237 (2008). 1238 12. Rittweger, J. Vibration as an exercise modality: how it may work, and what its 1239 potential might be. Eur. J. Appl. Physiol. 108, 877–904 (2010). 1240 13. Navarro, K. L. et al. Mouse anesthesia: The art and science. ILAR J. 62, 238–273 1241 (2021). 1242 14. Rao, S. S. Mechanical Vibrations . (Pearson, Upper Saddle River, NJ, 2010). 1243 15. Singleton, W. T. The Body at Work: Biological Ergonomics. (Cambridge University 1244 Press, Cambridge, England, 1983). 1245 16. Nikooyan, A. A. & Zadpoor, A. A. Mass–spring–damper modelli ng of the human 1246 body to study running and hopping – an overview. Proc Inst Mech Eng H 225, 1121–1247 1135 (2011). 1248 17. Karnopp, D. Active and semi-active vibration isolation. J. Vib. Acoust. 117, 177–185 1249 (1995). 1250 18. Lakie, M. & Campbell, K. S. Muscle thixotropy-where are we now? J. Appl. Physiol. 1251 126, 1790–1799 (2019). 1252 19. Huey, E. L. et al. The auditory midbrain mediates tactile vibration sensing. Cell 188, 1253 104-120.e18 (2025). 1254 20. Smith, S. D. & Kazarian, L. E . The effects of acceleration on the mechanical 1255 impedance response of a primate m odel exposed to sinusoidal vib ration. Ann. 1256 Biomed. Eng. 22, 78–87 (1994). 1257 21. Fairley, T. E. & Griffin, M. J . The apparent mass of the seated human body: vertical 1258 vibration. J. Biomech. 22, 81–94 (1989). 1259 22. Matsumoto, Y . & Griffin, M. J. Dynamic response of the standing human body 1260 exposed to vertical vibration: Influence of posture and vibration magnitude. J. Sound 1261 Vib. 212, 85–107 (1998). 1262 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 63 23. Mansfield, N. J. & Griffin, M. J. Non-linearities in appare nt mass and 1263 transmissibility during exposure to whole-body vertical vibrati on. J. Biomech. 33, 1264 933–941 (2000). 1265 24. Hinz, B. & Seidel, H. The nonlinearity of the human body’s dynamic response during 1266 sinusoidal whole body vibration. Ind. Health 25, 169–181 (1987). 1267 25. Wakeling, J. M. & Nigg, B. M. Modification of soft tissue v ibrations in the leg by 1268 muscular activity. J. Appl. Physiol. 90, 412–420 (2001). 1269 26. Tamura, Y ., Hatta, I., Matsuda, T., Sugi, H. & Tsuchiya, T. Changes in muscle 1270 stiffness during contraction recorded using ultrasonic waves. Nature 299, 631–633 1271 (1982). 1272 27. Hatta, I., Sugi, H. & Tamura, Y . Stiffness changes in frog skeletal muscle during 1273 contraction recorded using ultrasonic waves. J. Physiol. 403, 193–209 (1988). 1274 28. Lakie, M. & Robson, L. G. Thi xotropy: the effect of stimulation in frog muscle. Q. 1275 J. Exp. Physiol. 73, 627–630 (1988). 1276 29. Lännergren, J. The effect of low-level activation on the me chanical properties of 1277 isolated frog muscle fibers. J. Gen. Physiol. 58, 145–162 (1971). 1278 30. Dimitriou, M. Human muscle spindles are wired to function a s controllable signal-1279 processing devices. Elife 11, (2022). 1280 31. Marasco, P. D. & de Nooij, J. C. Proprioception: A New Era set in motion by 1281 emerging genetic and bionic strategies? Annu. Rev. Physiol. 85, 1–24 (2023). 1282 32. Wei, L. & Griffin, M. J. Mathematical models for the appare nt mass of the seated 1283 human body exposed to vertical vibration. J. Sound Vib. 212, 855–874 (1998). 1284 33. Shahabpoor, E., Pavic, A. & Racic, V . Identification of mass–spring–damper model 1285 of walking humans. Structures 5, 233–246 (2016). 1286 34. Karnopp, D., Crosby, M. J. & Harwood, R. A. Vibration contr ol using semi-active 1287 force generators. J. Eng. Ind. 96, 619–626 (1974). 1288 35. Proske, U. & Gandevia, S. C. The proprioceptive senses: their roles in signaling body 1289 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 64 shape, body position and movement, and muscle force. Physiol. Rev. 92, 1651–1697 1290 (2012). 1291 36. Wakeling, J. M., Nigg, B. M. & Rozitis, A. I. Muscle activi ty damps the soft tissue 1292 resonance that occurs in response to pulsed and continuous vibr ations. J. Appl. 1293 Physiol. 93, 1093–1103 (2002). 1294 37. Pollock, R. D., Woledge, R. C., Mills, K. R., Martin, F. C. & Newham, D. J. Muscle 1295 activity and acceleration during whole body vibration: effect o f frequency and 1296 amplitude. Clin. Biomech. (Bristol, Avon) 25, 840–846 (2010). 1297 38. Harazin, B. & Grzesik, J. THE TRANSMISSION OF VERTICAL WHOL E-BODY 1298 VIBRATION TO THE BODY SEGMENTS OF STANDING SUBJECTS. J. Sound 1299 Vib. 215, 775–787 (1998). 1300 39. Vidal, P.-P., Degallaix, L., Josset, P., Gasc, J.-P. & Cull en, K. E. Postural and 1301 locomotor control in normal and vestibularly deficient mice: Postural and locomotor 1302 control in mice. J. Physiol. 559, 625–638 (2004). 1303 40. Smith, B. J. H., Cullingford, L. & Usherwood, J. R. Identif ication of mouse gaits 1304 using a novel force-sensing exercise wheel. J. Appl. Physiol. 119, 704–718 (2015). 1305 41. Heglund, N. C., Cavagna, G. A. & Taylor, C. R. Energetics a nd mechanics of 1306 terrestrial locomotion. III. Ener gy changes of the centre of ma ss as a function of 1307 speed and body size in birds and mammals. J. Exp. Biol. 97, 41–56 (1982). 1308 42. Walter, R. M. Kinematics of 90 degrees running turns in wild mice. J. Exp. Biol. 206, 1309 1739–1749 (2003). 1310 43. Cregg, J. M. et al. Brainstem neurons that command mammalian locomotor 1311 asymmetries. Nat. Neurosci. 23, 730–740 (2020). 1312 44. Thuong, O. & Griffin, M. J. The vibration discomfort of sta nding persons: 0.5–16-1313 Hz fore-and-aft, lateral, and vertical vibration. J. Sound Vib. 330, 816–826 (2011). 1314 45. Holsgrove, T. P., Zeeman, M. E., Welch, W. C. & Winkelstein , B. A. Pain after 1315 whole-body vibration exposure is frequency dependent and indepe ndent of the 1316 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint 65 resonant frequency: Lessons from an in vivo rat model. J. Biomech. Eng. 142, (2020). 1317 46. Thuong, O. & Griffin, M. J. The vibration discomfort of sta nding people: relative 1318 importance of fore-and-aft, lateral, and vertical vibration. Appl. Ergon. 43, 902–908 1319 (2012). 1320 47. Valentine, T. Upside-down faces: a review of the effect of inversion upon face 1321 recognition. Br. J. Psychol. 79 ( Pt 4), 471–491 (1988). 1322 48. Ramachandran, V . S. Perception of shape from shading. Nature 331, 163–166 (1988). 1323 49. Adachi, I., Chou, D. P. & Hampt on, R. R. Thatcher effect in monkeys demonstrates 1324 conservation of face perception across primates. Curr. Biol. 19, 1270–1273 (2009). 1325 50. Wang, M.-Y . & Takeuchi, H. Individual recognition and the “ face inversion effect” 1326 in medaka fish (Oryzias latipes). Elife 6, (2017). 1327 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 13, 2025. ; https://doi.org/10.1101/2025.11.12.687975doi: bioRxiv preprint