Adaptation of the angular vestibulo-ocular reflex to head movements in rotating frames of reference

Abstract

Head movements in a rotating frame of reference are commonly encountered, but their long term effects on the angular vestibulo-ocular reflex (aVOR) are not well understood. To study this, monkeys were oscillated about a naso-occipital (roll) axis for several hours while rotating about a spatial vertical axis (roll-while-rotating, RWR). This induced oscillations in roll and pitch eye velocity and continuous horizontal (yaw) nystagmus. For several hours thereafter, simple roll in darkness induced horizontal nystagmus and pitch and roll oscillations. The rising and falling time constants of the horizontal velocity indicated that the nystagmus arose in velocity storage. The continuous nystagmus was correlated with a phase shift of vertical eye velocity from 90° to 0° re head position. As the phases reverted toward pre-adaptive values, the horizontal velocity declined. Similar yaw nystagmus and pitch and roll velocities were produced by oscillation in roll after adaptation with roll and horizontal optokinetic nystagmus (OKN), but not after adaptation with pitch-while-rotating (PWR). Findings were explained by a model that shifted the roll orientation vector of velocity storage toward the pitch axis during adaptation with RWR and Roll & OKN. This shift produced modulation in vertical eye velocity in the post adaptive state, which was approximately in phase with roll head position, generating horizontal nystagmus. Similar orientation changes to prolonged exposure to complex motion environments may be responsible for producing post-stimulus motion sickness and/or mal de debarquement.

Appendix

In this appendix, we review the canal and otolith afferent activation patterns during pure pitch and roll as well as during PWR and RWR obtained from previous studies (Raphan et al. 1999). We also review the orientation vector organization of velocity storage (Dai et al. 1991; Raphan and Sturm 1991; Raphan and Cohen 2002; Cohen and Raphan 2004; Raphan 2009).

Canal and otolith activation during pure pitch, PWR, pure roll, and RWR

Studies of pitch and roll while rotating have shown that intact vertical semicircular canals are critical for activating velocity storage to generate the continuous horizontal nystagmus during both pitch while rotating and roll while rotating (Raphan et al. 1983). This work suggested that there was a correlation between the phases of the oscillatory vertical canal and otolith activity related to head position that generated the steady state continuous horizontal eye velocity response (Raphan et al. 1983; Raphan et al. 1999). Consistent with this prediction, activity of vertical canal afferents was close to being in phase with otolith afferent activity during pitch and roll while rotating (Raphan et al. 1999). How the head position and afferent vertical canal modulation frequencies are related in the various conditions of this study is shown diagrammatically in Fig. 7A. During pure pitch, the anterior canal activity oscillates at 90° phase leading head position. The posterior canal activity is 180° out of phase with the anterior canal activity (Fig. 8Aa). During PWR, both anterior and posterior canal activations oscillate closer in phase with head position as the planes of the vertical canals become maximally activated when they move closer to the plane of the constant spatial yaw rotation (Fig. 8Ab). During pure roll (Fig. 8Ac), anterior and posterior canal activities are in phase and lead Left Side Down postion (LSD) by 90°. During RWR (Fig. 8Ad), anterior canal activity is in phase with Right Side Down Head Position (RSD) as it is maximally activated when the plane of the canal is close to the plane of the rotation. The posterior canal phase also shifts, but is 180° out of phase with the anterior canal activation.

Fig. 8

A Simulated modulation of semicircular canal activity during pure head pitch (a), PWR (b), pure head roll (c), and RWR (d). The simulations show Head Position (black trace), which is representative of otolith activation, as well as the modulation of the anterior and posterior canals for the left labyrinth. B The system matrix (H₀) and its orientation vectors (eigenvectors) when the head is upright (a–c), tilted about a roll axis (d–f) and tilted about a pitch axis (g–i). When upright (a), the system matrix is diagonal (b), and the orientation vectors are orthogonal along the roll, pitch, and yaw axes of the head (c). When the head is tilted about a roll axis (d), the system matrix has a yaw to pitch cross-coupling term (h_py). The yaw eigenvector that stays along the spatial vertical. Similarly, when the head is tilted about a pitch axis (g), the system matrix has a yaw to pitch cross-coupling term (h_ry) and the yaw eigenvector stays along the spatial vertical. The pitch and roll orientation vectors remain along the head coordinate frame regardless of head tilt

The post-adaptive data suggest that the Canal-Otolith Correlator, shown in the model of Fig. 7a), does not receive information directly from the vertical canals and otolith afferents. Rather, it utilizes activity from the pitch component of velocity storage, which introduces a 90° phase shift relative to otolith afferent activity rather than directly from the vertical canal afferents (Fig. 7a, x). The output of the correlator then drives the horizontal mode of velocity storage (Fig. 7a). The postulate that the correlator utilizes input from the vertical mode of velocity storage would also explain why horizontal velocity was generated during Roll, following adaptation to Roll & OKN.

The lateral canal afferent activated oscillated at twice the frequency of oscillation of the vertical canal and otolith activity. Because of the lack of correlation between the lateral canals and the other signals, they could not have contributed to the continuous horizontal nystagmus during PWR or RWR. They contributed a double frequency oscillation in the horizontal nystagmus (Raphan et al. 1999).

Orientation vectors of velocity storage

Velocity storage has been modeled as a three-dimensional integrator, which has a system matrtix, H₀, and an output represented by its state, x (Fig. 7a) (Dai et al. 1991; Raphan and Sturm 1991; Raphan and Cohen 2002; Cohen and Raphan 2004; Raphan 2009). Velocity storage is characterized by orientation vectors in a head coordinate frame that are represented by the eigenvectors of the system matrix (Dai et al. 1991; Raphan and Sturm 1991). Before adaptation, the system matrix H₀, is diagonal (Fig. 8Bb), corresponding to orientation vectors (eigenvectors) of velocity storage that are approximately orthogonal in the upright position and are aligned with cardinal axes of the head (Fig. 8Bb, c). When the head is tilted about a head roll axis (Fig. 8Bd), the system matrix is no longer diagonal, but now has a cross coupling term from yaw to pitch, h_py and a yaw orientation that is aligned close to the spatial vertical (Fig. 8Be, f). Similarly, tilts about the pitch axis (Fig. 8Bg) introduce a cross coupling term, h_ry and a yaw orientation vector that is close to the spatial vertical. These kinds of static tilts have the effect of orienting the yaw orientation vector to the spatial vertical (Fig. 8Bf, i), creating a non-orthogonal basis for velocity storage. A consequence of this, is that yaw stimulation generates pitch or roll eye velocity so that the vector of eye velocity tends to align with the yaw orientation vector, which is no longer along the head yaw (Dai et al. 1991; Raphan and Sturm 1991). The pitch and roll orientation vectors move with the head and are linked to the body coordinate frame. The present study considers how a simple adaptation of the roll orientation vector could explain the data on post-adaptation with RWR.