Abstract
Two recent models of the subjective vertical (SV), proposed by Dai, Curthoys and Halmagyi (1989; abbr. DCH-model) and by Mittelstaedt (1983a, b, 1988; abbr. M-model), respectively, are analyzed, experimentally tested and evaluated with regard to the role they attribute to the consequences of the pitchedup orientation of the labyrinth in the human skull. In the DCH-model the response to the resulting pitch shear is, after subtraction from a constant reference and normalization, multiplied with the response to roll shear and thence leads to the well-known Müller- and Aubert-deviations from veridicality. In the M-model the pitched-up orientation is accounted for by a shortcut type of coordinate transformation: subtraction of the saccular afference from the utricular response to pitch yields the head-fixedX-(pitch)component, whereas their addition yields the head-fixedZ-component. The SV will be veridical in a normal range of head positions if the amplitude of the saccular response is related to the utricular one as the sine is to the cosine of the pitched-up angle (≈ 30°). But then, typical deviations must result in other specifiable positions. The DCH- and the M-model are experimentally tested in positions where they predict SV-deviations of opposite sign. The results in 5 partly naive partly well-versed probands are highly significantly different from the DCH-model predictions whereas well compatible with those of the M-model. The implications for the models under scrutiny are discussed as well as for the global modelling of systems when essential internal constituents are unknown or inaccessible.