The disappearing hand: vestibular stimulation does not improve hand localisation

Background Bodily self-consciousness depends on the coherent integration of sensory information. In addition to visual and somatosensory information processing, vestibular contributions have been proposed and investigated. Vestibular information seems especially important for self-location, but remains difficult to study. Methods This randomised controlled experiment used the MIRAGE multisensory illusion box to induce a conflict between the visually- and proprioceptively-encoded position of one hand. Over time, the perceived location of the hand slowly shifts, due to the fact that proprioceptive input is progressively weighted more heavily than the visual input. We hypothesised that left cold caloric vestibular stimulation (CVS) augments this shift in hand localisation. Results The results from 24 healthy participants do not support our hypothesis: CVS had no effect on the estimations with which the perceived position of the hand shifted from the visually- to the proprioceptively-encoded position. Participants were more likely to report that their hand was ‘no longer there’ after CVS. Taken together, neither the physical nor the subjective data provide evidence for vestibular enhanced self-location.

119 the proprioceptively-encoded hand position, whereas the initial location and/or drift of the left 120 hand would be shifted toward the visually-encoded hand position). 121 Clarifying the potential role of the vestibular system in bodily self-consciousness is 122 important because vestibular stimulation has been proposed for treating a wide range of 123 conditions involving body awareness, despite a lack of strong supportive evidence (for critical 124 reviews see e.g. Grabherr et al., 2015;Miller, 2016). Findings from this study may help further 125 the understanding of the role of vestibular information in self-location, particularly regarding the 126 prioritisation of conflicting sensory input (i.e. visual vs. proprioceptive). This knowledge may 127 help explain underlying mechanisms responsible for observed vestibular induced effects in 128 patients with impaired bodily self-consciousness. 129 Taken together and despite some contradictory findings, we hypothesised that left cold 130 CVS augments the self-location component of bodily self-consciousness. This should be 131 reflected by a more pronounced shift from visual to proprioceptive location in the MIRAGE 132 illusion, such that vestibular stimulation compared to a sham condition leads to (i) a faster drift 133 in perceived location, toward the proprioceptively-encoded hand position, (ii) an initial perceived 134 location that is closer to the proprioceptively-encoded hand position, or (iii) both. An a priori power analysis was performed to determine the number of participants. Given 140 the use of a novel tool to assess hand localisation after CVS, it was difficult to calculate effect 141 sizes based on previous data. Therefore, we assumed a generic medium to large effect size of ƒ = 142 0.3 with a power = 0.8. Estimations were performed with MorePower 6.0 (Campbell & 143 Thompson, 2012) detecting a required sample of 22 participants. We recruited 24 to allow for a 2 144 x 2 counterbalanced design. This sample size is slightly bigger than previous studies testing the 145 influence of vestibular stimulation on bodily self-consciousness and body representation but 146 using different tasks (Lopez et al., 2010;Lopez, Schreyer, Preuss, & Mast, 2012). Participants 147 were recruited through noticeboards and social media. The mean age was 25 ± 9 SD; 14 were 148 female. One participant was replaced because no CVS response could be evoked (no nystagmus 149 and no reported subjective effects). All participants needed to be right-handed according to a 150 shortened version of the Edinburgh Handedness Inventory (Veale, 2014) and had normal or 151 corrected-to-normal vision. Participants were assessed with a short medical history questionnaire 152 and were excluded if they reported otological, neurological, or psychological problems. A 153 physician verified the integrity of the eardrum (otoscopy) before and after the experiment. 154 Participants received financial compensation for their participation.

155
The study was approved by the local ethics committee (University of South Australia, 156 Application ID: 32955) and was performed in accordance with the Declaration of Helsinki. 157 Written informed consent was obtained before commencing the experiment. Caloric vestibular stimulation (CVS) was performed with participants lying supine with 161 their head positioned 30˚ upright, in order to place the horizontal semicircular canal in a vertical 162 position. The left external ear canal was irrigated for 2 minutes with 100 ml of cold water (20˚ 163 C). Typically, left cold CVS elicits a slow-phase nystagmus towards the left side and a fast-phase 164 nystagmus towards the right side accompanied by an illusory feeling of self-motion to the right 165 and predominantly activates the right hemisphere, where the left hand is represented (Lopez, 166 Blanke, & Mast, 2012). Importantly, the right hemisphere is the dominant hemisphere for 167 vestibular function (Dieterich et al., 2003). Sham stimulation consisting of irrigating the left ear 168 with water at body temperature (37˚) was used as a control. The sham stimulation elicits similar 169 extra-vestibular cues (e.g. tactile stimulation), but is widely viewed as not leading to an 170 activation of the vestibular system . Two experimenters verified the 171 effectiveness of the stimulation by visually assessing the nystagmus. A vestibular stimulation 172 questionnaire (adapted from Lenggenhager, Lopez, & Blanke, 2008; Lopez et al., 2010) assessed 173 the subjective effects of vestibular stimulation. In particular, illusory motion was assessed on a 174 numerical rating scale from 0 to 6 (0 = no motion, 6 = very strong motion). 13 other common 175 symptoms associated with vestibular stimulation were also assessed on a numerical rating scale 176 from 0 to 6 (0 = absent, 6 = severe). These were general discomfort, nausea, vertigo, racing 177 heartbeat or palpitations, difficulty concentrating, drowsiness, faintness, sweating or cold sweat, 178 need to vomit, headache, fatigue, pallor, and blurred vision (please see Supplementary material, 179 Fig. S2a and b). The vestibular stimulation questionnaire was administered before (in order to 180 familiarise participants with the questionnaire and to clarify any clinical terms), and immediately 181 after CVS and sham stimulation.

182
In line with other experimental protocols, adaptation and self-location testing with the 183 MIRAGE system (see below) were conducted after the nystagmus had subsided, in order to 184 prevent perceptual biases induced by the directional beating of the nystagmus itself. This is 185 possible since CVS studies have revealed effects that outlast the duration of the stimulation as 186 shown by several neuroimaging and behavioural studies (please see Ferrè, Haggard, Bottini & 187 Iannetti, 2015, for a thorough discussion). Participants were able to rest for 30 minutes between 188 the two conditions, to minimise a possible carry-over effect. The MIRAGE system consists of a 2-chamber box with a double sided mirror between the 192 bottom chamber and the top chamber, and a video screen facing down placed on the ceiling of 193 the top chamber. The system is arranged so that video images of the lower chamber are projected 194 onto the mirror of the top chamber. Therefore, the mirror appears to show what is in the chamber 195 below. The result is that, when the participants look down towards their hands, they see an image 196 of their own hands at approximately the same location as the actual hands. Participants 197 comfortably sit in front of the box while a bib blocks the view of their arms or any other visual 238 the visual image of the hand (see Fig. 1). The experimenter then placed the participants' hands 239 on the floor of the box, where the hands remained until the first round of testing was complete.

240
Participants then performed the localisation task, by saying "stop" when the arrow aligned 241 with where they believed their middle finger to be. The localisation task was performed 6 times 242 at 15-second intervals, and the stop position of the arrow recorded for each trial. After 243 approximately every third trial, participants were verbally reminded to focus on where they felt 244 their middle finger to be. The actual physical location of the middle finger was recorded by the 245 experimenter. Participants then removed their hands from the MIRAGE box and completed the 246 questionnaire about subjective experience of the hand that was just tested. This process was then 247 repeated for the other hand. It is not known exactly how long the effects of CVS last, but prior 248 research using CVS to modulate bodily sensations found changes that persisted for at least 15 249 minutes after the CVS (Ferré et al., 2013). Consequently, the MIRAGE and localisation tasks 250 were performed as quickly as possible after the caloric or sham stimulation and always within 251 this 15-minute window. Testing of both hands was completed in an average of 10 minutes 36 252 seconds (±54 seconds) after completion of the CVS or sham; the maximum time required for 253 MIRAGE testing was 14 minutes from completion of the CVS or sham. Hand (right vs. left) and 254 condition (CVS vs. sham) were counterbalanced and randomised in blocks of four. After the first 255 session, participants were able to rest for 30 minutes. 256 Finally, at the end of the experiment, participants were asked five debriefing questions to 257 determine subjects' awareness of the illusion: 1. What do you think we were studying in the 258 experiment? 2. Did you notice a difference between the two stimulation conditions? 3. What do 259 you think was happening in the position tracking part (i.e. when the hands were between the two 260 blue bars) of the experiment? 4. Were you aware that you were moving your hands during the 261 position tracking part of the experiment? 5. If yes to previous question: In what direction were 262 your hands moving? 263 264 Statistical analysis 265 Hand location error was calculated as the difference between the participants' estimated 266 position and the last seen position of the middle finger. Last seen position was used rather than 267 the physical position because we were interested in initial localisation and speed of drift. To test 268 this, a linear regression was performed on location at the 6 time points, resulting in an intercept 269 (reflecting initial displacement relative to visual location) and slope (reflecting speed of drift 270 from visual to physical or proprioceptive location) for each hand and condition. Slope and 271 intercept were analysed using a repeated measures analysis of variance (ANOVA) with the 272 factors Stimulation (CVS, sham) and Hand (left hand, right hand) as within-subjects factors. 273 Responses to many of the MIRAGE questions were not normally distributed (and could not be 274 transformed to be normal) and were therefore analysed non-parametrically using Friedman Two    Hypothesis testing (Stimulation and Hand) showed that Question 4 ("It seemed like my 312 right/left hand was no longer there") was significant (p = .036), with post hoc Wilcoxon tests for 313 CVS vs. sham significant for the left hand (p = .012) but not the right hand (p = .039). None of 314 the other questions revealed a significant difference between CVS and sham. See Table 1 for the 315 statements of all questions and full statistical indices and please refer to Supplementary material 316 Fig. S1 illustrating the results of the questionnaire. To assess illusory motion due to CVS, we asked participants: "Did you perceive any type 324 of motion during the stimulation?" and, if they responded affirmative, they were questioned 325 further about how strong this perception was for the whole body, upper body, head, arms, and, 326 legs (please refer to the method section "caloric vestibular stimulation" for more details on the 327 questionnaire). All items were different between conditions (all p < .01), indicating that 328 participants -as expected -experienced stronger illusory motion perception after CVS than after 329 sham. Ratings were highest for head motion (mean 3.9 ± 2.2 SD) and lower for the arms (1.8 ± 330 2.0 SD) and legs (1.4 ± 1.9 SD). One participant did not report experiencing illusory motion, but 331 nystagmus was observed and other symptoms reported, so we were confident that CVS was 332 effective in this participant. All other participants reported the strength of illusory motion to be 2 333 or greater in the CVS condition. 334 We also asked participants to describe in what direction they perceived themselves to move 335 (perception of illusory own body-motion). While many participants reported experiencing a 336 circular motion to the right, a few participants found it difficult to clearly indicate the directional 337 quality or found it difficult to distinguish between illusory own-body motion and illusory motion 338 of the room (i.e. visual field). Despite some moderate general discomfort and other symptoms such as vertigo and 342 dizziness, all participants were able to complete the experiment. One participant reported slight 343 dizziness (between 1 and 2 on a scale from 0 -6) up to five days later; while unusual, this long 344 duration of effects was also observed once during a pilot test.

345
Participants frequently reported vestibular related side effects after CVS but seldom after 346 sham (Supplementary material Fig. S2a and b). Most symptoms (10/13), namely, general 347 discomfort, nausea, vertigo, racing heartbeat or palpitations, difficulty concentrating, drowsiness, 348 faintness, sweating or cold sweat, need to vomit and blurred vision, were rated significantly 349 higher (all p < .05) after CVS than after sham. Head tension or headache and fatigue were no 350 more common after CVS than after sham (p > .1); pallor was not significant (p = .07). Most 351 (>50%) participants reported general discomfort and difficulty concentrating, 50% reported 352 vertigo and fewer reported other symptoms (please refer to Supplementary material, Fig. S2a and  353 b).

Manuscript to be reviewed
Participants were not aware of the precise hypothesis tested but they all noticed a 357 difference between CVS and sham, typically pointing out that they experienced more and 358 stronger side effects during CVS than they did during sham (see section above). Nine 359 participants hinted that they believed that we were testing how CVS impairs their ability to 360 perform the task.

361
Most participants reported that their hands moved as needed to remain between the two 362 blue bars, but either said that they were moving only slightly back and forth, or inwards (i.e. 363 towards each other). Four participants reported that their hands were in fact moving outwards 364 (correct observation). We investigated whether caloric vestibular stimulation (CVS) changes the relative 368 weighting of proprioceptive and visual information in self-location, or bodily perceptions of the 369 hands, when participants receive conflicting visual and proprioceptive information regarding 370 hand position. Our results show that, although the feeling of the target hand (i.e. the hand that 371 was out of view and had to be localised) is different after CVS, as compared to after sham, there 372 was no effect on perceived location of the hand, nor on the weighting of proprioceptive and 373 visual information during self-location. These findings are evidenced by no differential effect of 374 condition on the slope or intercept of the location data, but a stronger reported feeling after CVS 375 that the hand was 'no longer there' (main effect question 4 of the MIRAGE questionnaire). Post 376 hoc testing showed that this main effect seems to be driven by the results of the left hand 377 (significant result for the left hand and a trend only for the right hand) possibly reflecting the 378 predominant activation of the right hemisphere after left cold CVS and/or the right hemisphere 379 dominance of the vestibular system.  Participants were in general not aware of the mismatch between the seen and the physical 393 position of the hand. For most participants, the MIRAGE illusion was effective in creating a 394 mismatch between visual and proprioceptive localisation of the hands; only four participants 395 were aware that their hands physically moved in a different direction from the visual image of 514 with late-onset of nystagmus or a weaker response, both of which could possibly affect our 515 results.

516
Our final consideration is that we did not lodge and lock our experimental protocol prior to 517 conducting this experiment. Doing so enhances the transparency of research and is now 518 recommended practice for observational and clinical designs in many fields and including this 519 step would have enhanced the confidence with which the reader could accept our results (e.g. 520 Lee et al., 2018). This investigation of whether CVS changes the relative weighting of proprioceptive and 524 visual information in self-location, or bodily perceptions of the hands, showed that although the 525 subjective feeling of the hands appeared to be disrupted by CVS, there is no effect on self-526 location of the hand, nor on the weighting of proprioceptive and visual information during self-527 location. The methodological advantages of the MIRAGE procedure, and the naivety of 528 participants to the illusory manipulation, add weight to the finding that CVS does not improve 529 self-location of a single body part.      Localisation of the middle finger, average for all participants at each of the 6 trials; standard deviation bars shown going up for CVS and down for sham. Slope and intercept were calculated based on these results. Data markers for left and right hands are shifted slightly left and right, respectively, of trial number to more clearly differentiate the data. Distance is measured from 0 = visual location to 13.65 cm = physical location of the middle finger.