Anchoring the Self to the Body: Vestibular Contribution to the Sense of Self

Method

We combined galvanic vestibular stimulation (GVS), an established method for controlled artificial vestibular stimulation, with the graphesthesia task. Eighteen healthy participants volunteered for the study (see Participants in the Supplemental Material available online). Participants kept their eyes closed during the trials. Carbon rubber electrodes were placed binaurally over each participant’s mastoid processes and fixed in place with adhesive tape. Electrodes for sham stimulation were placed on each participant’s neck in a left-anodal/right-cathodal configuration 5 cm below the location of the GVS electrodes. All electrodes were placed at the beginning of the session and remained in place throughout.

Each trial began with a 1-s pause, after which the participant received either 6 s of GVS (1-mA right-anodal/left-cathodal GVS or 1-mA left-anodal/right-cathodal GVS) or 6 s of a sham stimulation that was applied to the cervical area and did not affect the vestibular organs (see Methods in the Supplemental Material). One second after the stimulation began, a letter form (b, d, p, or q, chosen at random) was traced on the participant’s forehead with a cotton swab in a single motion beginning with the end of the letter stem (see Fig. 1a). After 4 s of stimulation, a tone signaled that the participant should name the letter. Three stimulation conditions (right-anodal/left-cathodal GVS, left-anodal/right-cathodal GVS, and sham stimulation; Fig. 1b) were administered in separate blocks in randomized order. Each block consisted of 32 trials, with eight repetitions of each letter, presented in random order. There was a 5-s break between trials. The experimenter was blind to the experimental hypothesis.

OPEN IN VIEWER

Fig. 1.

Experimental setup and results. On each trial, galvanic vestibular stimulation (GVS) or sham stimulation was delivered while the experimenter traced an ambiguous letter (b, d, p, or q) in the center of the participant’s forehead. The task was to name the letter. Sham stimulation, right-anodal/left-cathodal GVS, and left-anodal/right-cathodal GVS were delivered in separate blocks via electrodes placed over the mastoid processes (for GVS) or the sides of the neck (for sham stimulation; b). + = anode; – = cathode. The graph (c) shows the percentage of first-person-perspective responses as a function of condition. Error bars indicate standard errors of the mean. Asterisks indicate significant differences between sham stimulation and both GVS polarities (p < .05).

Results

Each response was classified as indicating a third-person perspective (e.g., b read as b), a first-person perspective (e.g., b read as d), or an error (any other response; e.g., b read as p). The percentage of first-person-perspective responses was then calculated for each stimulation condition. Most participants showed a mixture of first- and third-person perspectives (see Quality of Perspective Responses in the Supplemental Material). No order effects emerged (see Order-Effect Analysis in the Supplemental Material). An analysis of variance showed a significant main effect of stimulation condition, F(2, 34) = 4.625, p = .017. Follow-up testing revealed that, compared with sham stimulation, both right-anodal/left-cathodal GVS, t(17) = −2.652, p = .017, Cohen’s d = 0.52, and left-anodal/right-cathodal GVS, t(17) = −2.442, p = .026, Cohen’s d = 0.34, caused significantly more first-person-perspective judgments. No difference was found between right-anodal/left-cathodal GVS and left-anodal/right-cathodal GVS, t(17) = 0.961, p = .350 (Fig. 1c). These results cannot be explained by nonspecific factors, such as changes in arousal or distraction, because the percentage of error responses was similar across conditions (right-anodal/left-cathodal GVS: M = 4.86%, SEM = 1.91%; left-anodal/right-cathodal GVS: M = 7.06%, SEM = 2.37%; sham stimulation: M = 6.66 %, SEM = 2.34%; ps > .05 for all comparisons; see Nonspecific Effects of GVS: Analyses of Nonperspectival Errors in the Supplemental Material).

Discussion

Distortions of the relation between self and body in patients with vestibular disorders were first reported by Bonnier (1905). Vestibular stimulation has been shown previously to influence judgments about the orientation of visually presented images of bodies (Lenggenhager, Lopez, & Blanke, 2008). However, no previous studies have linked vestibular signals to the perspective taking in relating stimuli to one’s own body.

Graphesthetic perspective varies across skin regions (Corcoran, 1977; Parsons & Shimojo, 1987). Stimuli on the forehead, as used in this experiment, are normally, but not always, read from a first-person perspective, as if viewed from inside the head. Stimuli on the back of the head are normally read from a third-person perspective. The brief GVS we used promoted an “embodied eye” (Sekiyama, 1991): GVS, compared with sham stimulation, promoted more frequent first-person perspective relative to third-person perspective. Out-of-body experience (Blanke, 2012) would imply a bias in the opposite direction. If the self were located outside the body, stimuli on the skin would be seen with third-person perspective. Instead, we found bias toward first-person perspective in the GVS conditions. Our low-intensity GVS may have augmented a natural vestibular contribution to embodiment, rather than disrupting processes that normally anchor the self within the body. Indeed, GVS at the intensities we used produces functional postural responses similar to those produced by natural vestibular signals (Day et al., 1997). Disembodied experiences may require different stimulation patterns, stronger intensities, or direct stimulation of vestibular cortical areas rather than the vestibular periphery (Blanke et al., 2002).

The present study provides the first experimental evidence that vestibular inputs promote a first-person perspective. The temporoparietal junction might be involved in this process. This region receives multisensory body-related signals from somatosensory, proprioceptive, and vestibular pathways. Moreover, epileptic patients spontaneously reporting out-of-body experiences were found to have bilateral functional changes in temporoparietal regions (Blanke & Arzy, 2005). Our finding suggests that vestibular inputs contribute to the unity of multisensory experience that underlies self-awareness.