Here, we show the remarkable capacity of a subject with a severe somatosensory loss (including cervical information and cutaneous cues with the sitting platform) to maintain an active sitting posture (i.e. without back or side rests) in the absence of vision. We argue that the DS controlled her posture through vestibular signal processing. Compelling evidence for a vestibular-based postural control comes from the marked increase of postural oscillations when the DS's head was unconsciously deviated from its primary, visually perceived, straight-ahead position. In this situation, and in the lack of head-on-trunk position information, the vestibular signals did not provide the DS with veridical information about her trunk displacements in space. Processing vestibular signals with such undetected head-trunk configuration is likely to have led to inappropriate postural adjustments with respect to the actual body oscillations. For instance, with the head unconsciously rotated to the left, the DS most likely detected (either consciously or unconsciously) backward displacements when she was actually moving leftward. Such craniotopic updating of body displacement with neck proprioception deficit presumably resulted in series of inappropriate postural adjustments in response to the vestibular stimulations.
A great deal of the evidence for the contribution of vestibular signals to the control of posture comes from studies conducted with labyrinthine-defective subjects [e.g. [30
]] and from studies that tested the effects of external labyrinthine stimulations on healthy subjects (e.g. galvanic [7
] or caloric [32
] stimulations). While these studies indisputably attest the importance of vestibular information in postural control, the activation of the vestibular system resulting from such stimulations (or the lack of activation in the case of the patients) does not correspond to that normally arising from the tested postural tasks. In the present experiment, creating a mismatch between body motion and vestibular information by rotating the DS's head subliminally, proved to be an efficient way to demonstrate that the vestibular afferent signals generated by the body oscillations while seated were also contributive. Control subjects, for whom cervical afferent signals provide reliable information about head-trunk configuration, could transform the primary head-centered vestibular information into a trunk-centered frame of reference of body motion with the eyes closed [5
]. Consequently, their balance was not perturbed when their head was rotated toward a shoulder. Interestingly, the difference in body stability between the DS and controls almost completely vanished when vision was available. This was true whether or not the head was rotated. With vision, the DS could presumably refresh her body image according to the new, visually-perceived, orientation of the head relative to the trunk. This would allow her to estimate body motion through visuo-vestibular integration and to produce postural adjustments accordingly. With or without visual feedback, it is most likely that the control of posture required a great deal of attention for the DS [33
Despite her severe loss of neck muscle proprioception, the DS's postural responses to GVS were still mainly oriented towards the anode side. For safety reason and in order to preserve the confidence of the DS regarding the experimental procedures, we did not deliver GVS when the DS had the head unconsciously deviated towards a shoulder. The large lateral CoP shifts of the DS following GVS support previous findings obtained in individuals with somatosensory deficits [12
]. GVS had no or only marginal, and relatively insignificant, effect on the controls' CoP. This result also confirmed other studies that have employed GVS in sitting conditions [12
]. For control subjects, proprioception provided massive flow of information related to their actual body configuration and balance. Greater stimulations may be necessary to evoke larger postural responses, especially in conditions characterized by large surface of support, as in the present experiment.
Both the large CoP displacement observed when the DS had the head unconsciously deviated on the trunk and the large CoP shift after GVS onset suggest great sensitivity of the DS to vestibular stimulations. Such increased sensitivity has been reported when proprioceptive sense is deteriorated [12
] or when body stability is unsecured [30
]. Increased influence of vestibular information for the DS could result from the absence of the gating effect of proprioception on the vestibulospinal drive which is usually observed in control subjects [38
]. Also, response to vestibular stimulation may have been augmented by the high level of background muscular activity [40
] presumably present in the DS while seated without visual information.
An unexpected result of the present experiment was the DS's increased body stability for the GVS condition as compared to the condition without GVS (Head centered condition, eyes closed). Indeed, before and after the GVS-evoked large shift towards the anode side, the DS's CoP remained relatively stationary. Reasons for this improved balance in the GVS session remain uncertain. This experiment was the first attempt of the DS to sit without back or side supports since her deafferentation. It is then plausible that visual information, which was available in the conditions ran prior to the GVS condition (i.e. Head centered and Head rotated conditions with eyes open), allowed the DS to acquire a new strategy to control her sitting posture. This strategy may involve stiffening and freezing the trunk and legs, a strategy used by patients and elderly individuals with sensory impairments to control their balance while standing and walking [42
]. Of course, such a strategy would certainly prove to be inefficient for contexts where the posture would be perturbed externally [45
]. The lack of muscular activity recordings in the present experiment prevents verification of this hypothesis.