The present study investigated the temporal dynamics of ocular torsion and vertical vergence in terms of their relative onset times as triggered by visual, vestibular, and visuovestibular motion. Their respective response times were furthermore assessed with respect to the amount of visual information density, i.e. clutter, in a viewed optokinetic scene, and the acceleration of the stimulus. This allowed us to investigate if the aforementioned factors impacted the onset times of the two eye movements as well as their temporal dynamics in relation to one another. The decay rates of the torsional and vergence eye position following prolonged optokinetic rotations were also assessed.

The present study collated data retrieved from three separate trials investigating the eye movement responses of ocular torsion and vertical vergence to visual, vestibular, and visuovestibular motion. Results from these studies have been previously presented in terms of slow-phase gain of each eye movement. However, no data concerning the temporal dynamics of the torsional and vergence responses has been published. The present study made use of these datasets to allow a comprehensive statistical framework, allowing a higher statistical power than those retrieved for the isolated investigations.

Investigation One (Wibble and Pansell 2019) included 12 healthy subjects, 8 male and 4 female, with an average age of 43 years (ranging from 25 to 72). Thirteen individuals, comprising 7 men and 6 women with an average age of 25 years (ranging from 23 to 34), were recruited as volunteers for Investigation Two (Wibble et al. 2020a). Investigation three (Wibble et al. 2020b) included 16 healthy participants, 8 men and 8 women (ranging from 19 to 65 years of age). Subjects were recruited according to the same criteria for all investigations: None of the participants had any medical conditions or used drugs known to affect the central nervous system. All demonstrated normal or corrected visual acuity (VA) of at least 1.0 using the logarithm of the minimum angle of resolution (logMAR chart), stereoscopic vision of at least 200″ of arc (tested with the Lang II stereotest), and normal eye movement. Before participation, any considerable latent strabismus was ruled out using the cover test. None had a history of vertigo, and vestibular function was evaluated using a horizontal head impulse test, which showed no refixation saccade. Balance was assessed using the Romberg’s test on a soft platform. All participants could reliably fixate their eyes on a viewed target, and maintain gaze position in darkness, as evaluated using a head mounted eye-tracker (see eye and head movement recording).

Before enrolling in the study, all participants provided informed written consent after receiving detailed explanations about the study’s nature, along with both written and oral information about the procedure. The research adhered to the principles outlined in the Declaration of Helsinki and received approval from the Regional Ethics Committee of Stockholm (EPN 2018-1768-31-1).

Investigations one and two implemented a protocol evaluating torsional and vergence eye movements during visual, vestibular, and visuovestibular rotations in the roll plane; Investigation one implemented two different levels of visual information density in the viewed visual scene, i.e., visual clutter, while investigation two issues all stimulations at two different accelerations. These stimulations were carried out with durations of only a few seconds. By contrast, investigation three exposed subjects to visual rotations for a duration of 20 s, allowing a greater displacement of the eye position; these trials were repeated using the two different levels of visual clutter as first implemented in investigation one. Altogether, the present investigation collated the temporal dynamics retrieved from these three investigations, allowing us to explore how torsional and vergence latencies are affected by acceleration, visual clutter, and modalities, i.e., visual, vestibular, or visuovestibular motion. The sensory-specific protocols are outlined below, and can be found summarized in Fig. 1. All participants were exposed to all trials within each investigation.

The experimental setups of the investigations included in the present study. Investigation one and two featured visual, vestibular, and visuovestibular trials, whereas investigation three only featured prolonged optokinetic stimulations. The acceleration (acc) and amplitude (amp) of each stimulation are given in the corresponding column and row. The level of intensity (low or high) refers to the order of magnitude for each variable (visual clutter or acceleration level)

All visual stimulations were carried out on a projected screen (res 1024 × 768; contrast 2000:1; update frequency 60 Hz) by a front video projector (NEC NP-M350X, NEC Display Solutions Ltd., Tokyo) presented at an eye-screen distance of 200 cm. Visual elements consisted of white lines on a black background for investigations one and two, and black lines on a white background for investigation three, presenting comparable visual contrasts. The low intensity visual clutter featured 19 lines while the high intensity held 38 lines, all being viewed at a visual angle of 0.93 degrees. Investigation one and three used both these clutter levels, while investigation two used only the high intensity. These clutter levels always provided the same lighting levels during the stimulation, as tested using a photometer (universal photometer model S4; Hagner, Solna, Sweden).

Investigation one featured an optokinetic acceleration of 66 deg/s2 to an amplitude of 33 degrees, investigation two presented stimuli at 28 deg/s2 to an amplitude of 14 degrees and 58 deg/s2 to an amplitude of 28 degrees, and investigation three’s stimulation moved at 72 deg/s2 for 1440 degrees. Investigation one and two featured subjects seated, while subjects were standing during investigation three. Participants were tasked to always fixate on a high-contrast fixation point centred in the visual scene during each trial; these started with 20 s of baseline during which time the static visual scene was presented, after which the active motion was initiated. After the termination of motion, all subjects retained fixation for another 20 s until the recording was stopped. Investigations one and three featured optokinetic motion in both clockwise and counter-clockwise directions, but as no effect of direction was seen only counter-clockwise motion was implemented during investigation two.

All isolated vestibular stimulations in investigations one and two were carried out by rotating participants using an in-lab constructed mechanized sled which rotated on two belts powered by two AC Brushless Servo Motors (Baldor BSM90C, 400 V) in a dark room. Light pollution was precluded by framing the room with blackout curtains. To minimize darkness adaptation, participants were also exposed to an intensive bright light 5 s prior to the start of recording, after which a baseline was established during 20 s before the movement was implemented. Vestibular trials were only implemented during investigations one and two and were carried out at the same amplitudes but at half the acceleration of their respective visual counterpart. The centre-of-rotation was set to the glabella for all subjects in order to adjust for differences in height.

Visuovestibular trials were carried out according to the same principle as the vestibular stimulations, i.e., through whole-body rotations of each participant, but with subjects viewing the static visual scene associated with each respective investigation. This means that investigation one featured visuovestibular trials where subjects were rotated while viewing both low- and high intensity visual clutter. Investigation two instead introduced only the high intensity visual clutter but rotated at two different accelerations. This meant that subjects were exposed to the summated motions of the visual and vestibular trials of each investigation.

Eye movements were tracked using the head-mounted Chronos Eye Tracker Device (C-ETD; Chronos Inc, Berlin, Germany). This system was designed for binocular recordings, and the recording rate was 100 Hz with high spatial resolution of < 0.05° for horizontal and vertical eye movements, and < 0.1° for ocular torsion. The torsional movements were quantified as rotational displacements of the iris around the centre of the pupil by tracking iris features using the integrated Chronos software. This technique relied on template matching each frame with the initial reference frame obtained at the start of recording. Through cross-correlation, where a value of 1.0 indicated perfect matching, each frame was assigned a quality measure ranging from 0 to 1.0. Frames with a quality value below 0.5 were excluded from the analysis.

The horizontal and vertical displacements of the pupil were initially calibrated by instructing the participant to perform a series of eye movements to a pattern of dots with known separations, thereby allowing for the conversion of pupil displacement into angular degrees of horizontal, vertical, and torsional eye rotations. As video-based systems have been shown to be susceptible to iris occlusions, uneven lighting conditions, or poor pupil definition, multiple iris segments were collected for analysis, with each segment evaluated to determine the highest possible signal quality. Furthermore, all measurements were conducted under uniform lighting conditions, with the room darkened using blackout cloth and the projector serving as the sole light source. The tracked iris region was positioned away from any corneal light reflexes or lid shadows that could potentially interfere with the signal. In the event that the eye position signal was disrupted due to blinking during stimulation the trial was repeated.

A head tracking system was incorporated into the headgear for simultaneous recording of head movements in six dimensions (three rotational and three translational). This enabled precise measurement of head movements, ensuring that the subject remained stationary or moved at a precise rate in accordance with the vestibular or visual stimulation requirements. Head movements were monitored in all trials to ensure that no confounding activation of the VOR would influence the eye movement response. No such confounding head movement was seen for any trial.

The recorded sequences were subjected to processing using the analysis software integrated with the eye-tracking system (Chronos Vision GmbH, Berlin) to derive values for horizontal and vertical pupil positions, as well as torsional displacement of iris position, in degrees. The vertical skewing response was computed by subtracting the vertical eye position of the left eye from that of the right eye, while the torsional response was determined based on the eye with the most favourable signal-to-noise ratio. The retrieved data was plotted graphically in the Origin software (OriginPro 2017; OriginLab, Northampton, MA). An investigator manually evaluated all traces to determine the onset of both the torsional and vergence response. These were determined on the basis of presenting a clear deviation from the baseline in the direction corresponding to the movement stimulation. Examples of the torsional and vergence onsets can be found in Fig. 2. The decay rate of the eye position for investigation three was determined using the Origin software’s exponential decay function (ExpDecay1; y = y0 + A1*exp(−(x–x0)/t1)), and the time constant was retrieved. This was done for the time period immediately following the visual stimulation, i.e., between 40 and 60 s of the recording.

Representative traces illustrating the torsional and vergence responses at the start of visual, vestibular, and visuovestibular motion. The onsets of the vestibular and visuovestibular trial are highlighted by the head position indicating the stimulation start. The torsional and vergence onsets are indicated by the dotted lines

Stimulation triggers were performed manually, meaning that there was a margin of error for the onset of the stimulation itself; the stimulation and the eye-movement recording were triggered through different programs, and both were started by the same investigator simultaneously pressing two triggers. This created uncertainty in calculating the latencies for torsional and vergence eye movements in response to visual stimulations, as the start of visual motion could not be determined with high enough precision. Thanks to the had-mounted accelerometer, the vestibular and visuovestibular responses could be determined in relation to the start of the head movement with high temporal precision.

Statistical analyses were carried out according to two principles: (1) First level analyses were carried out within each investigation to allow for within-subject analyses where appropriate. These were carried out using repeated measures ANOVA in JASP (Version 0.16.4; JASP Team 2019, which was also used to perform test of normality. In the case of non-normal distribution or missing data points, a Generalized Linear Model (GLM) was implemented using SPSS Statistics 25 for Windows. (2) Data from investigations one and two were collated into one GLM, allowing a thorough evaluation of the onsets of ocular torsion and vertical divergence; note that investigation three was excluded from this model as it presented a different method of stimulation as outlined in Fig. 1. α was set to 0.05.