A common assumption of perceptual psychophysics is that the psychometric function is symmetric about a mean value. This assumption has been successfully challenged for contrast sensitivity (Burkhardt et al. 1984
; Garcia-Perez and Alcala-Quintana 2009
) and time perception (Wackermann and Spati 2006
). However, symmetry is still assumed in the experimental design and analysis of most psychophysical experiments (Treutwein 1995
; Leek 2001
; Macmillan and Creelman 2005
). In many prior vestibular perception studies, it has been assumed in both the design of the experiment and analysis of the responses that perceptual thresholds are similar in opposite directions of movement (Melvill Jones and Young 1978
; Benson et al. 1989
; Grabherr et al. 2008
; MacNeilage et al. 2010
Directional asymmetries in vestibular reflexes are the basis of clinically useful tests such as caloric responses (Bárány 1921
; Becker 1979
; Jacobson and Means 1985
), head thrust (Halmagyi and Curthoys 1988
), vertical axis rotation (Baloh et al. 1977
), and VEMP (Welgampola and Colebatch 2005
). Measurement of subjective visual vertical during static testing and off axis rotation has also been used to detect asymmetries in otolith function (Kingma 2006
). This testing measures the static response to a constant acceleration such as gravity.
Perceptual asymmetries may occur due to either bias or directional differences in dynamic sensitivity. Response bias is a common problem in forced choice experiments, and several methods of measuring it have been proposed (Macmillan and Creelman 2005
). A recent review of vestibular psychophysics has hypothesized that asymmetries in motion perception may be due to bias (Merfeld 2011
), but the only data cited are two subjects who were excluded from the author’s prior study and have not been published. Most calculations of response bias make an assumption that the underlying psychometric function is symmetric about the mean, which is not appropriate given the current evidence provided in this study. To measure response bias as directly as possible, we decided to measure the response in the absence of a stimulus. Preliminary results demonstrated that subjects were able to easily determine when no stimulus was delivered due to the absence of subtle vibration and sound cues that typically occurred with stimulus delivery. Subsequently, a null stimulus was designed that produced similar vibration and sound but no net motion. These results indicated that subjects tended to guess both directions about equally, and when bias was present, it was not associated with an asymmetry (Fig. ).
There is additional evidence that these asymmetries are not due to response bias alone. All subjects tended to have asymmetries only during a subset of test conditions. If these asymmetries were due to preferential guessing, they would be likely to occur in all test conditions for that subject. When subjects were tested with the same parameters in multiple sessions, ANOVA indicated that the day of testing accounted for a minimal amount of variation in the threshold determinations. Thus, thresholds even when asymmetric tended to be consistent across multiple days when the same condition was tested, so this phenomenon is unlikely due to chance (Figs. , , and ).
Finally, a bias implies that subjects perceive they are in motion even when they are stationary, which is, in essence, vertigo. Although this occurs pathologically, it is not commonly perceived in normal subjects.
In the current data, the BIC indicates that directional specific DS Guassian better explains the responses than a SS Guassian in three quarters of trial blocks, and the rate is even higher when the AIC is used. Even when a SS Guassian is used to fit data that is clearly asymmetric (Fig. ), the best fit includes only a small bias.
This paper presents a method for determining direction specific vestibular thresholds by using separate direction specific staircases and fitting separate psychometric functions to responses in each direction, which is novel to the vestibular field. These results indicate that vestibular perception thresholds are direction specific in a fraction of healthy people (Fig. ) and that, when present, asymmetries are consistently found during subsequent sessions (Figs. and ). However, when responses are averaged over the study population, most responses are symmetric (Fig. ). Interestingly, the one condition where a systematic asymmetry was present was for heave at 0.5 Hz, where most subjects were more sensitive to downward motion. This is a similar finding with a comparable marginal level of significance as previously reported using a different technique (Benson et al. 1986
A significant directional asymmetry in threshold was found 27% of the trial blocks and in at least one stimulus type in 92% of the subjects tested. These results indicate that such asymmetries are common and do not indicate vestibular pathology. The true incidence of directional asymmetries probably depends on how hard one looks for them. Increasing the number of stimulus presentations in each run and decreasing the step size in the staircase would allow more statistical power and perhaps demonstrate that smaller asymmetries are actually significant.
The origin of these perceptual asymmetries is unclear. Directionally asymmetric responses in the semicircular canals are well known as Ewald’s second law (Ewald 1892
). However, these asymmetries have been studied almost exclusively during high velocity rotation, well above the threshold of perception. In addition, the asymmetry in the vestibulo-ocular reflex is minimal for low velocity rotation (Baloh et al. 1977
; Katsarkas et al. 1995
Some evidence exists that the otolith-based vestibular reflexes may be asymmetric. The medial area of the utricle is larger than the lateral area in humans (Rosenhall 1972
) and three quarters of the neurons respond preferentially to ipsilateral tilt in monkeys (Fernandez and Goldberg 1976
). Transient linear motion produces a linear VOR (LVOR) which is commonly asymmetric at high accelerations in healthy humans (Lempert et al. 1998
; Crane et al. 2003
) and more asymmetric after an acute unilateral vestibular lesion, although symmetry returns to the normal range over time (Lempert et al. 1998
It is possible that the asymmetries seen in this study originate in the peripheral semicircular canals and otolith organs. Such asymmetries are likely well tolerated during daily activities because the vestibular system does not need a high degree of accuracy when responding to very low velocity stimuli such as those used in this study. It is likely that, when present, factors such as vision and proprioception play the predominate role in this domain.