Validating models of sensory conflict and perception for motion sickness prediction

Abstract

The human motion perception system has long been linked to motion sickness through state estimation conflict terms. However, to date, the extent to which available perception models are able to predict motion sickness, or which of the employed perceptual mechanisms are of most relevance to sickness prediction, has not been studied. In this study, the subjective vertical model, the multi-sensory observer model and the probabilistic particle filter model were all validated for their ability to predict motion perception and sickness, across a large set of motion paradigms of varying complexity from literature. It was found that even though the models provided a good match for the perception paradigms studied, they could not be made to capture the full range of motion sickness observations. The resolution of the gravito-inertial ambiguity has been identified to require further attention, as key model parameters selected to match perception data did not optimally match motion sickness data. Two additional mechanisms that may enable better future predictive models of sickness have, however, been identified. Firstly, active estimation of the magnitude of gravity appears to be instrumental for predicting motion sickness induced by vertical accelerations. Secondly, the model analysis showed that the influence of the semicircular canals on the somatogravic effect may explain the differences in the dynamics observed for motion sickness induced by vertical and horizontal plane accelerations.

Keywords: Motion sickness; Perceptual modelling; Sensory conflict; Sensory integration; State estimation.

Fig. 1

The hypothesized mechanism of velocity storage. The real angular velocity of the head ${\overrightarrow{\omega }}_{\text{h}}$ is measured by the semicircular canals. This measurement is imperfect and results in the high-pass filtering of the head angular velocity into ${\overrightarrow{\omega }}_{\text{hs}}$. This is compared with the output of the internally predicted semicircular canal output, ${\overrightarrow{\widehat{\omega }}}_{\text{hs}}$. The difference between the two is the conflict and is passed through the gain $K_{\omega c}$ to give an estimate of the real head angular velocity, ${\overrightarrow{\widehat{\omega }}}_{\text{h}}$

Fig. 2

The subjective vertical model. It is parameterized by: $K_{\text{ac}}$ is the acceleration feedback gain into the internal model in the form of an integral feedback $\frac{K_{\text{ac}}}{s}$, $K_{\text{gc}}$ is the gravity feedback gain into the internal model in the form of an integral feedback $\frac{K_{\text{ac}}}{s}$, $K_{\omega c}$ is the angular velocity conflict feedback gain into the internal model in the form of a proportional feedback SSC and $\overline{\mathit{SSC}}$ are first-order high-pass filters representing sensor dynamics of the semicircular canals, LP and $\overline{\mathit{LP}}$ are first-order low-pass filters capturing gravity estimation according to the Mayne equation

Fig. 3

The multi-sensory observer model. It is parameterized by: $K_{\text{a}}$, which multiplies the otolith magnitude conflict to create the internally estimated acceleration ${\overrightarrow{\widehat{a}}}_{\text{h}}$, $K_f$, which multiplies the otolith angle conflict to create the angular velocity computed from the angular difference between the internally estimated and sensed gravito-inertial force ${\overrightarrow{\widehat{f}}}_{\text{h}}$ and ${\overrightarrow{f}}_{\text{h}}$, $K_{f\omega }$, which has the same function but instead of being used to compute a new gravity estimate it is summed to create an internal estimate of angular velocity ${\overrightarrow{\widehat{\omega }}}_{\text{h}}$, and $K_{\omega }$, which multiplies the angular velocity conflict. The subsequent signal is summed with the estimate of angular velocity coming from $K_{f\omega }$. SSC and $\overline{\mathit{SSC}}$ are first-order high-pass filters denoting the semicircular canals and the internal model of the semicircular canals, respectively, and lastly, OTO and $\overline{\mathit{OTO}}$ are unit transfer functions denoting the otoliths and the internal model of the otoliths, respectively

Fig. 4

The particle filter model. It is parameterized by: the weighting factor $w_t^i$ given by a multiplication of Gaussian priors (of mean zero and variance ${\sigma }_A$ and ${\sigma }_{\omega }$) on the inertial acceleration $P({\overrightarrow{A}}_t)$ and angular velocity $P({\omega }_t)$, the head-to-canal transformation matrix $T_{\text{can}}^{-1}$, the integration time step $\delta t$ and the canal time constant $T_{\text{c}}$

Fig. 5

Perception predictions for EVAR, centrifugation and OVAR, according to the PFM (purple lines), SVM (dashed blue) and MSOM (dashed yellow), plotted against experimental data (red lines). The first row shows yaw velocity perception for EVAR. The second row shows roll perception during centrifugation. In the third and fourth rows, the normalized translation velocity perception (obtained by integrating translational acceleration) and rotation perception are shown for OVAR, respectively. The left column shows fits based on parameters obtained from the literature; the centre column perception-tuned fits; the right column sickness-tuned fits. All motion paradigms are performed in darkness

Fig. 6

Frequency-domain model responses of perceived motion. a responses of the perception-tuned model state estimates in response to small acceleration input whilst earth vertically orientated in darkness. The three lines represent SVM (blue), MSOM (yellow) and PFM (magenta). Input head acceleration is $a_h$, estimated head acceleration is ${\widehat{a}}_{\text{h}}$, estimated gravity is ${\widehat{g}}_{\text{h}}$ the units for both are ms${}^{-2}$. b responses of the sickness-tuned model state estimates

Fig. 7

Frequency sensitivity of motion sickness to sickening stimuli. a Perception-tuned frequency sensitivity to horizontal plane accelerations. b Perception-tuned frequency sensitivity to vertical plane accelerations. c Sickness-tuned frequency sensitivity to horizontal plane accelerations. d Sickness-tuned frequency sensitivity to vertical accelerations. The red line is experimentally observed frequency sensitivity, blue line is SVM, yellow line is MSOM and magenta line is PFM. The PFM is absent in b, d because it does not have any conflicts in the vertical direction

Fig. 8

SVM gravity conflict ${\overrightarrow{c}}_{\text{g}}$ for PRP (purple line) and LTA (dashed blue line) with respect to angular velocity feedback gain $K_{\omega }$

Fig. 9

Magnitude of cumulative conflict for five motion paradigms when normalized against CCCP for the three perception models. The left column shows results for the SVM (in blue); the middle columns show results for the MSOM (yellow); and the right column shows the results for the PFM (magenta). The top row a shows predicted sickness for perception-tuned models; the bottom row b shows the predicted sickness for sickness-tuned models. Red diamonds mark experimental results

Fig. 10

Kinematic chain used for complex motion generation