Effects of prolonged roll-tilt on the subjective visual and haptic vertical in healthy human subjects

Abstract

BACKGROUND:

While verticality perception is normally accurate when upright, a systematic bias (“post-tilt bias”) is seen after prolonged roll-tilt. The source of the bias could either be central (shifting “null” position) or related to changes in torsional eye-position.

OBJECTIVE:

To study the mechanisms of the post-tilt bias in vision-dependent and vision-independent paradigms and to characterize the impact of optokinetic stimulation.

METHODS:

The subjective visual-vertical (SVV) and subjective haptic-vertical (SHV) were measured after static roll-tilt (±90deg ear-down (“adaptation”) position; duration = 5 min; n = 9 subjects). To assess the effect of visual stimuli, a control condition (darkness) was compared with an optokinetic stimulus (clockwise/counter-clockwise rotation, 60deg/sec) during adaptation.

RESULTS:

A significant post-tilt bias was more frequent for the SVV than the SHV (72% vs. 54%, p = 0.007) with shifts pointing towards or away from the adaptation position with similar frequency. Exponential-decay time-constants were comparable for both paradigms and directions of shifts. The optokinetic stimulus had no effect on the bias for either paradigm.

CONCLUSIONS:

Emerging in both vision-dependent and vision-independent paradigms, the results support the hypothesis that the post-tilt bias results from a shift in the internal estimate of direction of gravity, while optokinetic nystagmus seems not to be a major contributor.

Keywords

Perception multisensory integration subjective visual vertical subjective haptic vertical vestibular

1 Introduction

Accurate and precise estimates of the direction of gravity are essential for spatial orientation and navigation (see [8] for review). Multiple sensory input systems including the otolith organs, the semicircular canals, vision and skin pressure sensors contribute to human verticality perception [1], with their relative contributions varying depending on their reliability. Bayesian optimal observer theory proposes a mechanism where the human brain combines all available sensory cues in a weighted fashion according to their relative reliabilities and prior likelihood to generate an internal estimate of the direction of gravity [3 , 19]. For verticality perception, the subject’s previous spatial orientation relative to gravity has a major impact. After prolonged (i.e., at least several minutes duration) static whole-body roll-tilt, a shift of the perceived direction of gravity (termed “post-tilt bias” [2]) either towards or (less frequently) away from the previous roll-tilted position has been observed for the subjective visual vertical (SVV) [11 , 25]. It has been postulated that this shift is the result of the brain incorrectly interpreting the position the subject is oriented to during prolonged roll-tilt as the “new upright” position, leading to a shift of the internal graviceptive null position [18].

Likewise, optokinetic rotatory stimuli are very powerful in biasing verticality perception as shown for the SVV [4]. Such shifts are roll-angle dependent and increase significantly with roll-tilt [27]. However, when using other, vision-independent paradigms to assess internal estimates of direction of gravity, effects of optokinetic stimulation vary significantly. Specifically, asking subjects to align a bar in complete darkness along perceived vertical (a task referred to as the subjective haptic vertical or SHV) at the time a rotatory optokinetic stimulus is presented, induces minor to non-significant shifts only [5]. Thus, optokinetic biases are likely paradigm-dependent, but emerging circular vection induced by prolonged optokinetic stimulation may contribute as well [5].

We have previously shown for the SVV that prolonged roll-tilt results in a post-tilt bias that exponentially decays with a median time constant of 71 seconds [18]. We therefore predict a post-tilt bias as well for other, vision-independent paradigms such as the SHV [16]. Alternatively, lacking evidence for a post-tilt bias in vision-independent paradigms would favor a task-specific origin of the post-tilt bias seen for the SVV. With regards to optokinetic rotatory stimuli, mixed results have been reported in the literature, with stimulus duration presentation likely being critical for the effect size. Whereas very brief stimulus presentation did not result in significant optokinetic-induced biases for the SHV [5], more prolonged (i.e. more than 5–10 sec) stimulus duration was linked to significant shifts in the subjective postural vertical in another study [4]. We therefore hypothesized that prolonged optokinetic stimulation results in an additional adaptational modulation of the post-tilt bias that outlasts the stimulus duration, either enhancing or reducing the effect size and that this effect again is relatively task-independent but requires a minimal stimulus duration of 5–10 seconds.

Based on these predictions we defined the study design, using two different setups (being either vision-dependent or vision-independent) to measure perceived direction of gravity, and exposed participants to two different stimuli that potentially bias verticality perception during the adaptation period (prolonged static roll-tilt and optokinetic rotatory stimuli).

2 Materials and methods

2.1 Study subjects and ethics statement

Nine healthy human subjects (2 females; aged between 22 and 42 years) were included in this study. All participants completed both paradigms (SVV and SHV). All participants were right-handed and adjustments were always done with the (dominant) right hand. Written informed consent of all subjects was obtained after a full explanation of the experimental procedure. The protocol was approved by the local ethics committee (Cantonal Ethics committee Zurich, BASEC 2016-00023) and was in accordance with the ethical standards laid down in the 2013 Declaration of Helsinki for research involving human beings.

2.2 Experimental setup

For all measurements, the participants were seated on a three-axis motor-driven turntable (Acutronic, Jona, Switzerland) and they were secured with a four-point safety belt. The head was restrained in a natural straight-ahead position using a thermoplastic mask (Sinmed, Reeuwyk, The Netherlands). Body movements were minimized by placing deflatable pillows around the shoulders, hips and legs. As the otolith organs, which are critical for graviception, are situated in the head, the subjects’ orientation in the roll plane will be referred as head-roll orientation, although roll movements of the turntable were whole body. The roll axis of the motorized turntable corresponded to the naso-occipital line passing between the subject’s eyes. The optokinetic rotatory stimulus was projected onto a sphere, 1.5 m from the subject, using a video projector mounted on the turntable. The rotating optokinetic stimulus was generated with the Psychophysics Toolbox [1, 19] and GNU Octave (version 3.2.3), and was made of randomly placed white dots on a black background. Turntable, arrow and bar orientation signals were stored on a computer hard disk after they have been digitized at 200 Hz for further analysis.

2.3 Experimental paradigms

For both the SVV and the SHV paradigm, baseline trials over a duration of five minutes were collected in upright position at the beginning of each recording session (see Fig. 1 for details). Then, subjects were roll-tilted and remained stationary in an adaptation position of either 90° right-ear down (RED) or 90° left-ear down (LED) roll orientation for five minutes. During this adaptation period, they were kept in darkness (“no optokinetic stimulation” condition) or were presented an optokinetic stimulus rotating (velocity = 60°/sec) either into the clockwise (“optokinetic CW” condition) or counter-clockwise (“optokinetic CCW” condition) direction for the entire adaptation period. This optokinetic roll velocity value was chosen based on previous observations showing a marked shift in SVV both when upright and roll-tilted using a velocity of 60°/sec Noteworthy, there was no optokinetic stimulation in the post-tilt period while the participants performed the adjustments as we were focusing on after-effects of gravitational and optokinetic stimuli on verticality perception.

Fig. 1

Illustration of required tasks at baseline (panel A), during the adaptation period (panel B) and immediately afterwards (panel C). For all three steps the duration was set to five minutes. Panel A: repetitive baseline arrow (SVV) or rod (SHV) (for illustrative purposes always an arrow is shown) adjustments along perceived vertical in whole-body upright position are collected. δ represents the deviation of the adjustments relative to earth-vertical, thus for perfect adjustments along earth-vertical. Panel B: during the adaptation period, subjects remain in a static roll-tilted position (referred to as α, set to±90°, only+90° shown for illustrative purposes) either in darkness (“no optokinetic stimulation” condition) or while watching an optokinetic stimulus that is rotating in either clockwise (“optokinetic CW” condition) or counter-clockwise (“optokinetic CCW” condition) direction. No arrow or bar adjustments are performed during this period. Panel C: Immediately after returning back upright, subjects again repetitively adjust a luminous arrow (SVV) or a rod (SHV) along perceived vertical in darkness (i.e., no optokinetic stimulus is shown). For both steps A and C errors in line/bar adjustments as reflected by angle δ are calculated.

The different turntable roll orientations were reached by movements with 10°/s² constant acceleration and deceleration. Importantly, the detection thresholds of semicircular canal stimulation [5, 22] and self-motion perception [15] are clearly exceeded by these acceleration/deceleration values. Thus, they could potentially have affected adjustment performance [10, 18]. Therefore, post-tilt SVV and SHV adjustments started five seconds after the turntable had reached its final (upright) position, as after such a delay post-rotatory torsional ocular drift at the time subjects confirm arrow adjustments was found to be small [20]. An acoustic signal indicated the start of all trials. To avoid training effects, starting arrow and bar roll orientations were random (allowing whole numbers only), thus, the distribution of roll starting angles was distinct for each subject and experimental session.

Paradigm specific aspects of the experimental settings are provided below. For each adaptation position and condition repetitive SVV and SHV trials were collected for five minutes in the immediate post-tilt period. Before starting measurements, subjects practiced both paradigms until they were able to perform adjustments reliably within the time limit.

2.3.1 Subjective visual vertical (SVV)

For the SVV paradigm, a red laser arrow (length 500 mm; width 3 mm; covering the central 9.5° of the binocular visual field) was projected on the surface of a sphere located 1.5 meters in front of the subject’s eyes. Subjects were allowed to wear corrective glasses/lenses. Subjects used a remote-control knob placed on a safety bar in front to rotate the arrow (spatial resolution = 0.1°) such that it was along the perceived direction of gravity and to confirm adjustments by pressing a button placed next to it. A time limit was set to five seconds for each trial and trials not completed within this period were discarded. The arrow disappeared after each trial, and reappeared two seconds later for the next trial. Starting orientation of the arrow was random within the entire roll plane for the SVV (i.e., ranging from 0 to 359°).

2.3.2 Subjective haptic vertical (SHV)

For the SHV paradigm, a tactile device (a tube, 290 mm long and 25 mm in diameter, spatial resolution = 0.04°) mounted on the safety bar in front of the subjects was used. The starting position of the bar was restricted to random CW or CCW offsets of 28 to 84° relative to the subject’s roll orientation due to the physical constraints of the human wrist. Subjects were instructed to grasp the rod using a wrap grip in darkness [7, 23] at the beginning of each trial (indicated by an acoustic signal) and to reassure themselves that the polarity of the rod was correct, i.e., that a short piece of Velcro attached to the bar was close to the thumb and not to the little finger. Trials started only after the subject had identified and correctly grasped the tactile device placed in front in darkness. Again, trial duration was limited to five seconds and trials not completed within this period were discarded. After pressing a button to confirm the final bar position, subjects released the bar, and during the two second inter-trial interval the bar was moved to the next starting position.

2.4 Definition of terms frequently used

Clockwise shifts relative to the earth-vertical axis (as seen by the subject) have positive signs and counter-clockwise shifts have negative signs. In the following, we will use the term trial-to-trial variability when we refer to the within subject standard deviation (SD). In relation to trial-to-trial variability, the term precision reflects the inverse, i.e. the degree of reproducibility. Furthermore, accuracy is defined as the magnitude of the mean adjustment error in a given paradigm.

2.5 Data analysis

Data was extracted and sorted according to the whole-body roll orientation and the condition for each participant using interactive programs written in Matlab 2017b (The MathWorks, Natick, MA, USA). Differences in adjustment errors and variability values for baseline trials and post-adaptation trials were calculated for both paradigms. Mean values (±1SD) were used when pooling individual data points as our data was normally distributed (tested at the level of individual trial conditions using the Jarque-Bera hypothesis test of composite normality, jbtest.m, Matlab 2017b). A generalized linear model (GLM) using SPSS 25 (IBM, Armonk, NY, USA) was applied for all statistical analyses if not specified otherwise. Main effects included the trial condition (n = 3; optokinetic off vs. optokinetic CW vs. optokinetic CCW), the direction of rotation of the arrow or the rod (n = 2; CW vs. CCW) and the adaptation position (n = 2, ±90deg roll-tilt). We kept the level of significance at a p-value of 0.05, and Fisher’s least significant difference (LSD) method was used to correct for multiple comparisons when using the GLM.

To identify tilt-adaptation induced shifts in accuracy, all SVV and SHV post-tilt trials were detrended, i.e., offset and drift from baseline trials was removed as previously described [18]. Post-tilt trials in both paradigms were fitted with an exponential decay function and a linear function and goodness-of-fit (R²-value) was compared using a parametric analysis of variance (ANOVA, Matlab 2017b). $y = a + b * x$ (1) $y = a * e^{\frac{x}{- Tc}} + c$ (2)

The decay time constant (Tc) was derived from those trials with significant drift, (i.e., with significant (p < 0.05) exponential fits), with separate analysis for post-tilt trials with initial offsets towards the previous roll-tilted position and those for initial offsets away from the adaptation position. In addition, to quantify the post-tilt bias in the SVV and the SHV paradigm, initial post-tilt offsets and drift amplitude based on the exponential fits were calculated (using the detrended traces).

3 Results

3.1 SVV paradigm

A mean (±1 SD) of 55±4.4 trials were collected within the five-minute post-tilt periods over the entire study population. Overall, an average of 3.4 (±4.0; 1SD) trials were discarded in each subject, reflecting a fraction of all trials ranging between 0 and 3.4%. Average trial time (all subjects and conditions pooled) was 2.9±0.4 sec and independent from the trial condition (as shown by statistical analysis using a GLM, with no significant main effects or interactions) (as shown in Table 1). Single subject raw data (subject #1) is presented in Fig. 2A. In this subject, the baseline measurement showed an initial offset that remained constant during the trial. This was in contrast to the post-tilt conditions, where drift was frequent.

Table 1
trial duration and number of trials (in square brackets) in both paradigms (always mean±1SD), reporting number of subjects in brackets

SVV paradigm

90°LED no OKN 90°RED no OKN 90°LED OKN CW 90°RED OKN CW 90°LED OKN CCW 90°RED OKN CCW

All subjects (n = 9) 3.0±0.3 [54±4] 2.9±0.4 [55±4] 2.9±0.4 [55±4] 2.8±0.3 [56±4] 2.9±0.4 [56±4] 2.9±0.4 [55±5]

Only those with initial offset towards the previous roll-tilted position and significant decay 3.3±0 [52±0] (n = 1) 2.8±0.3 [56±4] (n = 5) 3.8±0 [48±0] (n = 1) 2.7±0.3 [57±4] (n = 5) 2.9±0.5 [56±5] (n = 4) 2.9±0.2 [54±4] (n = 5)

Only those with initial offset away from the previous roll-tilted position and significant decay 3.1±0.2 [53±2] (n = 6) 3.1±0.4 [54±5] (n = 4) 2.9±0.1 [56±1] (n = 3) 3.2±0.3 [52±4] (n = 2) 3.0±0.0 [55±1] (n = 2) 3.6±0.0 [45±0] (n = 1)

Only those with no initial offset and no significant decay 2.4±0.1 [61±1] (n = 2) NaN (n = 0) 2.8±0.4 [56±5] (n = 5) 2.7±0.1 [58±1] (n = 2) 2.7±0.3 [56±5] (n = 3) 2.6±0.3 [59±3] (n = 3)

SHV paradigm (only trials with CW bar rotation)

90°LED no OKN 90°RED no OKN 90°LED OKN CW 90°RED OKN CW 90°LED OKN CCW 90°RED OKN CCW

All subjects (n = 9) 2.0±0.8 [30±7] 1.7±0.6 [33±6] 2.2±0.9 [29±7] 1.9±1.1 [32±8] 1.9±0.9 [31±6] 1.9±1.1 [33±9]

Only those with initial offset towards the previous roll-tilted position and significant decay 2.3±0.5 [28±4] (n = 3) 2.2±0 [29±0] (n = 1) 1.8±0.4 [33±4] (n = 2) 3.3±0 [23±0] (n = 1) 1.9±0.7 [28±3] (n = 2) 1.7±0.9 [34±7] (n = 4)

Only those with initial offset away from the previous roll-tilted position and significant decay 3.0±0 [33±0] (n = 1) 1.7±0.3 [33±4] (n = 2) 2.1±0.9 [31±8] (n = 5) 1.6±1.0 [30±1] (n = 2) 3.2±0 [23±0] (n = 1) 1.3±1.5 [38±13] (n = 3)

Only those with no initial offset and no significant decay 2.0±0.8 [28±9] (n = 5) 1.6±0.7 [34±7] (n = 6) 2.7±1.6 [22±6] (n = 2) 1.8±1.2 [34±9] (n = 6) 1.7±0.8 [33±7] (n = 6) 2.8±1.4 [22±13] (n = 2)

SHV paradigm (only trials with CCW bar rotation)

90°LED no OKN 90°RED no OKN 90°LED OKN CW 90°RED OKN CW 90°LED OKN CCW 90°RED OKN CCW

All subjects (n = 9) 2.1±0.9 [29±9] 1.8±0.7 [32±9] 2.2±0.8 [30±7] 1.9±1.1 [35±9] 1.9±0.9 [29±10] 1.9±0.9 [30±5]

Only those with initial offset towards the previous roll-tilted position and significant decay 1.5±0 [32±0] (n = 1) 1.7±0.8 [36±9] (n = 5) NaN (n = 0) 3.3±0 [38±9] (n = 6) 2.0±0.6 [31±4] (n = 2) 1.7±0.9 [32±7] (n = 4)

Only those with initial offset away from the previous roll-tilted position and significant decay 2.0±1.0 [32±14] (n = 3) 2.0±0.8 [27±9] (n = 3) 2.2±0.6 [35±7] (n = 3) 1.6±1.0 [33±0] (n = 1) 1.2±0.8 [39±5] (n = 3) NaN (n = 0)

Only those with no initial offset and no significant decay 2.3±0.9 [27±8] (n = 5) 2.1±0.9 [28±6] (n = 6) 1.8±1.2 [28±11] (n = 2) 2.4±1.0 [22±9] (n = 4) 2.1±1.0 [29±4] (n = 5)

SVV paradigm
	90°LED no OKN	90°RED no OKN	90°LED OKN CW	90°RED OKN CW	90°LED OKN CCW	90°RED OKN CCW
All subjects (n = 9)	3.0±0.3 [54±4]	2.9±0.4 [55±4]	2.9±0.4 [55±4]	2.8±0.3 [56±4]	2.9±0.4 [56±4]	2.9±0.4 [55±5]
Only those with initial offset towards the previous roll-tilted position and significant decay	3.3±0 [52±0] (n = 1)	2.8±0.3 [56±4] (n = 5)	3.8±0 [48±0] (n = 1)	2.7±0.3 [57±4] (n = 5)	2.9±0.5 [56±5] (n = 4)	2.9±0.2 [54±4] (n = 5)
Only those with initial offset away from the previous roll-tilted position and significant decay	3.1±0.2 [53±2] (n = 6)	3.1±0.4 [54±5] (n = 4)	2.9±0.1 [56±1] (n = 3)	3.2±0.3 [52±4] (n = 2)	3.0±0.0 [55±1] (n = 2)	3.6±0.0 [45±0] (n = 1)
Only those with no initial offset and no significant decay	2.4±0.1 [61±1] (n = 2)	NaN (n = 0)	2.8±0.4 [56±5] (n = 5)	2.7±0.1 [58±1] (n = 2)	2.7±0.3 [56±5] (n = 3)	2.6±0.3 [59±3] (n = 3)
SHV paradigm (only trials with CW bar rotation)
	90°LED no OKN	90°RED no OKN	90°LED OKN CW	90°RED OKN CW	90°LED OKN CCW	90°RED OKN CCW
All subjects (n = 9)	2.0±0.8 [30±7]	1.7±0.6 [33±6]	2.2±0.9 [29±7]	1.9±1.1 [32±8]	1.9±0.9 [31±6]	1.9±1.1 [33±9]
Only those with initial offset towards the previous roll-tilted position and significant decay	2.3±0.5 [28±4] (n = 3)	2.2±0 [29±0] (n = 1)	1.8±0.4 [33±4] (n = 2)	3.3±0 [23±0] (n = 1)	1.9±0.7 [28±3] (n = 2)	1.7±0.9 [34±7] (n = 4)
Only those with initial offset away from the previous roll-tilted position and significant decay	3.0±0 [33±0] (n = 1)	1.7±0.3 [33±4] (n = 2)	2.1±0.9 [31±8] (n = 5)	1.6±1.0 [30±1] (n = 2)	3.2±0 [23±0] (n = 1)	1.3±1.5 [38±13] (n = 3)
Only those with no initial offset and no significant decay	2.0±0.8 [28±9] (n = 5)	1.6±0.7 [34±7] (n = 6)	2.7±1.6 [22±6] (n = 2)	1.8±1.2 [34±9] (n = 6)	1.7±0.8 [33±7] (n = 6)	2.8±1.4 [22±13] (n = 2)
SHV paradigm (only trials with CCW bar rotation)
	90°LED no OKN	90°RED no OKN	90°LED OKN CW	90°RED OKN CW	90°LED OKN CCW	90°RED OKN CCW
All subjects (n = 9)	2.1±0.9 [29±9]	1.8±0.7 [32±9]	2.2±0.8 [30±7]	1.9±1.1 [35±9]	1.9±0.9 [29±10]	1.9±0.9 [30±5]
Only those with initial offset towards the previous roll-tilted position and significant decay	1.5±0 [32±0] (n = 1)	1.7±0.8 [36±9] (n = 5)	NaN (n = 0)	3.3±0 [38±9] (n = 6)	2.0±0.6 [31±4] (n = 2)	1.7±0.9 [32±7] (n = 4)
Only those with initial offset away from the previous roll-tilted position and significant decay	2.0±1.0 [32±14] (n = 3)	2.0±0.8 [27±9] (n = 3)	2.2±0.6 [35±7] (n = 3)	1.6±1.0 [33±0] (n = 1)	1.2±0.8 [39±5] (n = 3)	NaN (n = 0)
Only those with no initial offset and no significant decay	2.3±0.9 [27±8] (n = 5)	2.1±0.9 [28±6] (n = 6)	1.8±1.2 [28±11] (n = 2)	2.4±1.0 [22±9] (n = 4)	2.1±1.0 [29±4] (n = 5)

Fig. 2

For both the baseline measurements (indicated by a light gray background) and the post-tilt conditions, individual adjustments relative to earth-vertical are plotted against time in a single subject (#1). While in panel A results for the SVV paradigm are shown, panel B demonstrates the adjustment errors for the SHV paradigm, with trials with CW (in grey) and CCW (in black) bar rotation shown separately. The dashed black line indicates earth vertical position, i.e. accurate adjustments. Insets indicate the subjects roll orientation during the adjustments and the previous roll-tilted position. Abbreviations: LED = left-ear-down; RED = right-ear-down.

In a next step average adjustment errors (±1SD) over the five-minute periods were calculated for all subjects and trial conditions. Direction of arrow rotation had no effect on the adjustment errors (p = 0.391, paired t-test, Matlab 2017b), thus trials with CW and CCW arrow rotation were pooled for further analysis. The pattern of adjustment errors (Fig. 3A) and trial-to-trial variability (Fig. 3B) was very similar across the different conditions. This was later confirmed by statistical analysis as shown further below.

Fig. 3

Overall average (±1SD) adjustment errors (SVV = panel A; SHV = panel C) and trial-to-trial variability (SVV = panel B; SHV = panel D) are shown for both baseline and post-tilt conditions. For the SHV trials with CW (grey circles) and CCW (black circles) bar rotation were analyzed separately. For the post-tilt trials the specific adaptation condition (either 90° left-ear down (LED) or 90° right-ear-down (RED) and the visual background (no optokinetic stimulus, optokinetic CW, optokinetic CCW) is illustrated. Note that in the post-tilt period all trials were collected in darkness, i.e., no optokinetic stimuli were shown.

3.1.1 Linear fit vs. exponential fit

Goodness of fit (i.e., the R²-value) was significantly better (ANOVA, Matlab 2017b) when using the exponential fit compared to the linear fit in all post-tilt trial conditions for the SVV paradigm (0.27±0.21 vs. 0.21±0.19; df = 1; F-value = 14.59; p < 0.001). Thus, we used the exponential decay function for further post-tilt drift analyses.

We then grouped single trial conditions based on their drift properties, distinguishing trials with significant drift (as determined by the exponential decay function) and initial offset towards the adaptation position from trials with significant drift and initial offset away from the adaptation position and from trials with non-significant drift (or increasing error over time as noted in few trials). For illustrative purposes, individual data points were assigned to one of 16 equally long bins (covering 18.8 sec each), and mean (±1SD) values were calculated. As shown in Figure 4, these three patterns varied in frequency for the different trial conditions. Overall, an initial offset with significant drift was found in 39 out of 54 runs (72.2%). While a larger fraction of trials showed a post-tilt bias towards the previous roll-tilted position compared to away from this position (21 vs. 18), these differences were not significant (Fisher’s exact test, p = 0.689). For those traces with a significant exponential post-tilt decay, the time constant was determined. Time constants were calculated separately for those traces with initial offset towards the previous roll-titled orientation (mean (for all post-tilt conditions)±1SD = 74±67 sec) and for those traces with initial shifts away from the previous roll-tilted position (149±73 sec).

Fig. 4

Illustration of perceived vertical upon return to upright position after prolonged roll-tilt for the SVV paradigm. Runs are displayed in black for the overall mean (±1SD) and in grey for the single subject traces. For determining the initial offset and the decay characteristics, an exponential decay function was fitted (dashed line) to each trial (see methods section for details). Runs that were initially biased towards the previous roll orientation and had significant exponential decay are shown in the left column, runs that were biased away from the previous roll orientation and had significant exponential decay are presented in the middle column. Trials without a significant bias or significantly increasing offsets are illustrated in the right column. Each line refers to a different adaptation condition.

3.1.2 Initial post-tilt offsets and drift amplitudes –effect of the different trial conditions

For the different post-tilt conditions, average initial offsets ranged between –0.7±3.0° and 2.3±3.0°, whereas the drift amplitudes were between –1.9±2.4° and 0.8±2.6° (see Table 2 for details). Statistical analysis (GLM) of the initial post-tilt bias (i.e., the offset) for the SVV paradigm demonstrated no main effects for the condition (df = 2, chi-square = 1.414, p = 0.493) and the adaptation position (df = 1, chi-square=0.075, p = 0.784), whereas a significant interaction was found between these two parameters (df = 2, chi-square = 7.338, p = 0.026). Pairwise comparisons, however, demonstrated significant differences only for RED vs. LED (p = 0.023) in the no optokinetic stimulus condition and for LED in the no optokinetic stimulus vs. optokinetic CCW condition (p = 0.010). Likewise, statistical analysis of the drift amplitudes (obtained in order to compare the drift size amongst the different conditions) illustrated no main effects for the conditions (df = 2, chi-square = 0.410, p = 0.815) and the adaptation positions (df = 1, chi-square = 0.357, p = 0.550). Furthermore, no significant interactions between these parameters were noted (df = 2, chi-square = 1.783, p = 0.410).

Table 2
Drift characteristics after sustained roll-tilt for the SVV paradigm

Post-tilt conditions (all subjects (n = 9), always mean±1SD)

90°LED no OKN 90°RED no OKN 90°LED OKN CW 90°RED OKN CW 90°LED OKN CCW 90°RED OKN CCW

Initial offset [°] 2.3±3.0 –0.4±1.4 0.8±2.8 1.2±2.2 –0.7±3.0 1.0±3.2

Drift amplitude [°/5 min] –1.9±2.4 0.0±2.6 0.0±2.6 –1.9±2.8 0.8±2.6 –1.7±1.9

Post-tilt conditions (in subjects with initial offset towards the previous roll-tilted position and significant decay)

90°LED no OKN (n = 1) 90°RED no OKN (n = 5) 90°LED OKN CW (n = 1) 90°RED OKN CW (n = 5) 90°LED OKN CCW (n = 4) 90°RED OKN CCW (n = 5) All post-tilt conditions with significant decay (n = 21)

Initial offset [°] –1.4±0 0.7±0.2 –4.6±0 2.5±0.8 –3.3±1.9 3.1±0.9

Drift amplitude [°/5 min] 2.9±0 –2.1±0.4 5.3±0 –3.7±1.6 3.2±1.4 –2.9±0.8

Tc of exp. Decay [sec] 74±67

Post-tilt conditions (in subjects with initial offset away from the previous roll-tilted position and significant decay)

90°LED no OKN (n = 6) 90°RED no OKN (n = 4) 90°LED OKN CW (n = 3) 90°RED OKN CW (n = 2) 90°LED OKN CCW (n = 2) 90°RED OKN CCW (n = 1) All post-tilt conditions with significant decay (n = 18)

Initial offset [°] 3.9±2.2 –1.6±1.1 3.2±1.5 –2.3±1.8 2.3±2.5 (n = 2) –4.8±0 (n = 1)

Drift amplitude [°/5 min] –3.0±1.5 2.7±0.8 –2.5±1.1 2.2±1.0 –2.6±0.7 2.2±0

Tc of exp. decay [sec] 149±73

Post-tilt conditions (all subjects (n = 9), always mean±1SD)
	90°LED no OKN	90°RED no OKN	90°LED OKN CW	90°RED OKN CW	90°LED OKN CCW	90°RED OKN CCW
Initial offset [°]	2.3±3.0	–0.4±1.4	0.8±2.8	1.2±2.2	–0.7±3.0	1.0±3.2
Drift amplitude [°/5 min]	–1.9±2.4	0.0±2.6	0.0±2.6	–1.9±2.8	0.8±2.6	–1.7±1.9
Post-tilt conditions (in subjects with initial offset towards the previous roll-tilted position and significant decay)
	90°LED no OKN (n = 1)	90°RED no OKN (n = 5)	90°LED OKN CW (n = 1)	90°RED OKN CW (n = 5)	90°LED OKN CCW (n = 4)	90°RED OKN CCW (n = 5)	All post-tilt conditions with significant decay (n = 21)
Initial offset [°]	–1.4±0	0.7±0.2	–4.6±0	2.5±0.8	–3.3±1.9	3.1±0.9
Drift amplitude [°/5 min]	2.9±0	–2.1±0.4	5.3±0	–3.7±1.6	3.2±1.4	–2.9±0.8
Tc of exp. Decay [sec]							74±67
Post-tilt conditions (in subjects with initial offset away from the previous roll-tilted position and significant decay)
	90°LED no OKN (n = 6)	90°RED no OKN (n = 4)	90°LED OKN CW (n = 3)	90°RED OKN CW (n = 2)	90°LED OKN CCW (n = 2)	90°RED OKN CCW (n = 1)	All post-tilt conditions with significant decay (n = 18)
Initial offset [°]	3.9±2.2	–1.6±1.1	3.2±1.5	–2.3±1.8	2.3±2.5 (n = 2)	–4.8±0 (n = 1)
Drift amplitude [°/5 min]	–3.0±1.5	2.7±0.8	–2.5±1.1	2.2±1.0	–2.6±0.7	2.2±0
Tc of exp. decay [sec]							149±73

3.2 SHV paradigm

A mean (±1SD) of 61±13.7 haptic alignments were collected within the five-minute periods in all subjects. Overall, an average of 6.3 (±10.8; 1SD) trials were discarded in each subject, with a maximum of 32 trials in one subject where due (at least partially) to technical difficulties when the button triggering mechanism failed repetitively. These missed trials reflected a fraction of all trials ranging between 0 and 8.4%. Average trial time (all subjects and conditions pooled) was 2.0±0.8 sec and independent from the trial condition (as shown by statistical analysis using a GLM, with no significant main effects or interactions) (as shown in Table 1). Single subject raw data (subject #1) is presented in Fig. 2B. Noteworthy, trials with CW and CCW bar rotation differed in their offset; this was consistent between the different conditions. Amongst the different paradigms, the offset was fairly stable and no consistent drift was present (except for the 90° RED condition and CCW optokinetic stimulation).

The pattern of adjustment errors (Fig. 3C) and trial-to-trial variability (Fig. 3D) were very similar across the different conditions, but showed a clear effect of direction of bar rotation for the adjustment errors. Specifically, trials with CW bar rotation were accurate, whereas trials with CCW bar rotation fell short, resulting in a CW deviation relative to earth-vertical and relative to trials with CW bar rotations (Fig. 3C). This effect was confirmed by statistical analysis (CW vs. CCW, p < 0.001, paired t-test, Matlab 2017b). Thus, a separate analysis of trials depending on their direction of bar rotation was required.

3.2.1 Linear fit vs. exponential fit

Goodness of fit was significantly better when using the exponential fit compared to the linear fit in all post-tilt trial conditions for the SHV paradigm, thus again we used the exponential decay function for further analyses. This was true both for conditions with a CW bar rotation (0.25±0.20 vs. 0.18±0.17; df = 1; F-value = 19.26; p < 0.001) and for conditions with a CCW bar rotation (0.30±0.21 vs. 0.20±0.17; df = 1, F-value = 23.94; p < 0.001).

We grouped single post-tilt trial conditions based on their drift properties and initial offsets. As shown in Figs. 5 (CW bar rotations) and 6 (CCW bar rotations) these three patterns varied in frequency for the different trial conditions. Overall, a significant post-tilt bias was found in 27/54 (50.0%) and in 31/54 (57.4%) runs for trials with CW and CCW bar rotation, respectively. While for CW bar rotations trials with an offset away from the previous roll-tilted position were slightly more frequent than trials with a post-tilt bias towards the preceding roll-tilted position (14 vs. 13), the opposite was true for CW bar rotations (13 vs. 18; offset away from vs. offset towards previous roll-tilted position). Pooling trials with CW and CCW bar rotation, there was no significant difference in the frequency of shifts either towards or away from the previous adaptation position (Fisher’s exact test, p = 0.443).

Fig. 5

Illustration of perceived vertical upon return to upright position after prolonged roll-tilt for trials with CW bar rotation for the SHV paradigm. Runs are displayed in black for the overall mean (±1SD) and in grey for the single subject traces. For determining the initial offset and the decay characteristics, an exponential decay function was fitted (dashed line) to each trial (see methods section for details). Runs that were initially biased towards the previous roll orientation and had significant exponential decay are shown in the left column, runs that were biased away from the previous roll orientation and had significant exponential decay are presented in the middle column. Trials without a significant bias or significantly increasing offsets are illustrated in the right column. Each line refers to a different adaptation condition.

Fig. 6

Illustration of perceived vertical upon return to upright position after prolonged roll-tilt for trials with CCW bar rotation for the SHV paradigm. Runs are displayed in black for the overall mean (±1SD) and in grey for the single subject traces. For determining the initial offset and the decay characteristics, an exponential decay function was fitted (dashed line) to each trial (see methods section for details). Runs that were initially biased towards the previous roll orientation and had significant exponential decay are shown in the left column, runs that were biased away from the previous roll orientation and had significant exponential decay are presented in the middle column. Trials without a significant bias or significantly increasing offsets are illustrated in the right column. Each line refers to a different adaptation condition.

Time constants were calculated, again distinguishing the direction of the bar rotation. Overall, Tcs were shorter for those traces with initial offset towards the previous roll-titled orientation (CW mean (for all post-tilt conditions)±1SD = 82±80 sec; CCW = 73±74 sec) compared to those traces with initial shifts away from the previous roll-tilted position (CW = 117±78 sec; CCW = 115±82 sec).

Compared to the SVV paradigm, the fraction of runs resulting in a significant post-tilt bias was significantly smaller for the SHV paradigm (72.2% vs. 53.7%, Fisher’s exact test, p = 0.007).

3.2.2 Initial post-tilt offsets and drift amplitudes –effect of the different trial conditions

For the different post-tilt conditions and CW bar rotations, average initial offsets ranged between –0.9±2.8° and 2.2±6.8°, whereas the drift amplitudes were between –1.2±6.1° and 1.6±4.1° (see Table 3 for details). For the different post-tilt conditions and CCW bar rotations, average initial offsets ranged between 0.9±6.8° and 4.4±5.6°, whereas the drift amplitudes were between –6.2±5.4° and –1.7±7.1° (see Table 4 for details).

Table 3
Drift characteristics after sustained roll-tilt for the SHV paradigm and trials with CW bar rotation

Post-tilt conditions for trials with CW bar rotation (all subjects (n = 9), always mean±1SD)

90°LED no OKN 90°RED no OKN 90°LED OKN CW 90°RED OKN CW 90°LED OKN CCW 90°RED OKN CCW

Initial offset [°] –0.1±3.5 –0.9±2.8 1.7±6.6 –0.3±4.2 0.4±3.6 2.2±6.8

Drift amplitude [°/5 min] 1.6±4.1 1.0±3.9 –0.8±7.1 0.3±3.8 0.3±4.3 –1.2±6.2

Post-tilt conditions for trials with CW bar rotation (in subjects with initial offset towards the previous roll-tilted position and significant decay)

90°LED no OKN (n = 3) 90°RED no OKN (n = 1) 90°LED OKN CW (n = 2) 90°RED OKN CW (n = 1) 90°LED OKN CCW (n = 2) 90°RED OKN CCW (n = 4) All post-tilt conditions with significant decay (n = 13)

Initial offset [°] –3.0±0.5 0.9±0.0 –8.3±6.1 9.3±0 –2.4±0.7 8.3±3.9

Drift amplitude [°/5 min] 6.1±1.6 –6.0±0.0 9.5±6.4 –7.0±0 4.8±2.6 –6.2±1.4

Tc of exp. decay [sec] 82±80

Post-tilt conditions for trials with CW bar rotation (in subjects with initial offset away from the previous roll-tilted position and significant decay)

90°LED no OKN (n = 1) 90°LED no OKN (n = 2) 90°RED OKN CW (n = 5) 90°LED OKN CW (n = 2) 90°RED OKN CCW (n = 1) 90°LED OKN CCW (n = 3) All post-tilt conditions with significant decay (n = 14)

Initial offset [°] 7.2±0.0 –3.5±0.7 6.2±1.4 –4.4±1.7 6.0±0 –4.7±2.9

Drift amplitude [°/5 min] –5.6±0.0 5.8±0.2 –5.7±1.7 4.4±0.5 –5.9±0 6.5±1.7

Tc of exp. decay [sec] 117±78

Post-tilt conditions for trials with CW bar rotation (all subjects (n = 9), always mean±1SD)
	90°LED no OKN	90°RED no OKN	90°LED OKN CW	90°RED OKN CW	90°LED OKN CCW	90°RED OKN CCW
Initial offset [°]	–0.1±3.5	–0.9±2.8	1.7±6.6	–0.3±4.2	0.4±3.6	2.2±6.8
Drift amplitude [°/5 min]	1.6±4.1	1.0±3.9	–0.8±7.1	0.3±3.8	0.3±4.3	–1.2±6.2
Post-tilt conditions for trials with CW bar rotation (in subjects with initial offset towards the previous roll-tilted position and significant decay)
	90°LED no OKN (n = 3)	90°RED no OKN (n = 1)	90°LED OKN CW (n = 2)	90°RED OKN CW (n = 1)	90°LED OKN CCW (n = 2)	90°RED OKN CCW (n = 4)	All post-tilt conditions with significant decay (n = 13)
Initial offset [°]	–3.0±0.5	0.9±0.0	–8.3±6.1	9.3±0	–2.4±0.7	8.3±3.9
Drift amplitude [°/5 min]	6.1±1.6	–6.0±0.0	9.5±6.4	–7.0±0	4.8±2.6	–6.2±1.4
Tc of exp. decay [sec]							82±80
Post-tilt conditions for trials with CW bar rotation (in subjects with initial offset away from the previous roll-tilted position and significant decay)
	90°LED no OKN (n = 1)	90°LED no OKN (n = 2)	90°RED OKN CW (n = 5)	90°LED OKN CW (n = 2)	90°RED OKN CCW (n = 1)	90°LED OKN CCW (n = 3)	All post-tilt conditions with significant decay (n = 14)
Initial offset [°]	7.2±0.0	–3.5±0.7	6.2±1.4	–4.4±1.7	6.0±0	–4.7±2.9
Drift amplitude [°/5 min]	–5.6±0.0	5.8±0.2	–5.7±1.7	4.4±0.5	–5.9±0	6.5±1.7
Tc of exp. decay [sec]							117±78

Table 4

Drift characteristics after sustained roll-tilt for the SHV paradigm and trials with CCW bar rotation

Post-tilt conditions for trials with CCW bar rotation (all subjects, always mean±1SD)
	90°LED no OKN	90°RED no OKN	90°LED OKN CW	90°RED OKN CW	90°LED OKN CCW	90°RED OKN CCW
Initial offset [°]	3.9±5.5	0.9±6.8	2.8±4.8	3.8±6.5	2.0±5.1	4.4±5.6
Drift amplitude [°/5 min]	–3.1±5.0	–1.7±7.1	–2.8±4.8	–5.2±5.8	–2.2±5.5	–6.2±5.4
Post-tilt conditions for trials with CCW bar rotation (in subjects with initial offset towards the previous roll-tilted position and significant decay)
	90°LED no OKN (n = 1)	90°RED no OKN (n = 5)	90°LED OKN CW (n = 0)	90°RED OKN CW (n = 6)	90°LED OKN CCW (n = 2)	90°RED OKN CCW (n = 4)	All post-tilt conditions with significant decay (n = 18)
Initial offset [°]	–3.0±0.0	6.0±4.5	NaN	7.6±4.1	–3.7±0.3	9.9±1.8
Drift amplitude [°/5 min]	6.1±0.0	–7.2±2.8	NaN	–8.5±3.7	6.3±2.6	–11.2±3.8
Tc of exp. decay [sec]							73±74
Post-tilt conditions for trials with CCW bar rotation (in subjects with initial offset away from the previous roll-tilted position and significant decay)
	90°LED no OKN (n = 3)	90°RED no OKN (n = 3)	90°LED OKN CW (n = 3)	90°RED OKN CW (n = 1)	90°LED OKN CCW (n = 3)	90°RED OKN CCW (n = 0)	All post-tilt conditions with significant decay (n = 13)
Initial offset [°]	7.2±4.4	–5.4±1.2	7.3±2.5	–3.2±0	6.2±5.1	NaN
Drift amplitude [°/5 min]	–6.9±3.5	6.5±1.8	–7.4±2.1	2.1±0	–7.1±2.5	NaN
Tc of exp. decay [sec]							115±82

Statistical analysis (GLM) of the initial post-tilt bias for the SHV paradigm demonstrated no main effects for the condition (df = 2, chi-square = 1.390, p = 0.499) and the adaptation position (df = 1, chi-square = 0.008, p = 0.929), whereas a significant main effect for the direction of bar rotation (df = 1, chi-square = 6.655, p = 0.010) was noted. No significant 2-way or 3-way interactions were found, with p-values equal to 0.237 or larger.

Regarding the drift amplitudes, only a main effect for the direction of bar rotation (df = 1, chi-square = 6.610, p = 0.010) was found, with drift amplitude being significantly larger for trials with CCW bar rotation than for trials with CW bar rotation. Noteworthy, there were no main effects for the conditions (df = 2, chi-square = 0.728, p = 0.695) and the adaptation positions (df = 1, chi-square = 0.927 p = 0.336). Also, no significant 2-way or 3-way interactions were found, with p-values equal to 0.098 or larger.

4 Discussion

Here we asked whether the previously shown bias in verticality perception after prolonged static roll [3] is independent of the paradigm used and to which extent rotatory optokinetic stimuli are able to enhance or inhibit this post-tilt bias. After prolonged static roll-tilt, we noted a significant post-tilt bias in a majority of trials for both paradigms, with exponential-decay time-constants being comparable between the different conditions. Furthermore, no effect of optokinetic rotatory stimuli on the post-tilt bias were noted in either the SVV or the SHV paradigm, suggesting that such visual stimuli have no effect on the internal estimate of the direction of gravity outlasting the period of stimulus presentation.

4.1 The frequency and distribution of the post-tilt bias

After five minutes of static whole-body roll-tilt at±90° ear-down position, the fraction of trials with a significant post-tilt bias was significantly larger for the SVV paradigm than for the SHV paradigm (72.2% vs. 53.7%, p = 0.007). However, with a majority of trials showing a significant post-tilt bias in both vision-dependent (SVV) and vision-independent (SHV) paradigms, our results support the hypothesis that the post-tilt bias is the result of a shift in the internal estimate of direction of gravity [18]. Regarding the difference in the frequency of that such post-tilt biases with significant decay in the SVV and the SHV paradigm, several reasons may be considered. First, the size of the shift and its decay characteristics may be paradigm specific, with a more pronounced effect for the SVV. Second, the SHV has previously been shown to have a larger trial-to-trial variability near upright position than the SVV [16], potentially explaining why exponential decay function fitting was less successful in the SHV paradigm.

In contrast to our previous SVV study that found the post-tilt bias to deviate more often towards the adaptation position than away from it [18], deviations to either side were not significantly different in frequency in both paradigms for our current study. These discrepancies are likely related to the multiple factors contributing to the individual direction and size of drift in perceived vertical including dependence of serial adjustments over time [21] and emphasize that inter-individual differences in post-tilt bias are larger than previously thought. At the same time, however, we found no clear evidence for a major impact of trial duration or the fraction of discarded trials on our results. Specifically, there were no significant differences in trial duration throughout the different conditions as shown by statistical analysis (GLM, being true both for the SVV and the SHV paradigm) and neither the average trial duration times not the average number of trials included were systematically different for the distinct drift directions (towards vs. away from the previous adaptation position).

4.2 Decay characteristics of the post-tilt bias in the SVV and SHV paradigm

The exponential decay Tc for the post-tilt bias was similar for both paradigms, with average values ranging between 73 sec and 82 sec (trials with initial shifts towards the previous adaptation position) and between 115 sec and 149 sec (trials with initial shifts away from the previous roll-tilted position). Whereas for the SVV paradigm the average decay Tc was about twice as high for trials with an initial post-tilt bias away from the adaptation position compared to trials with an initial bias towards the adaptation position, such a marked difference was not noted for the SHV paradigm. The reasons for these paradigm-specific differences remain unclear. While a shift towards the previous roll-tilt position has been proposed to reflect a shift in the perceived upright position, i.e., suggesting that the current (roll-tilted) position is considered the “new upright” position, the mechanism in those subjects showing a shift away from the adaptation position is likely different. This may explain the discrepancies in decay Tc. However, whether the direction of these shifts is subject-specific or rather random needs to be determined in future studies. Noteworthy, the mean decay Tc for trials with an initial bias towards the previous adaptation position in our current SVV and SHV data was very similar to the median SVV decay Tc previously published by our study group [18], all being in the range of 71 sec to 82 sec.

Both observations support the hypothesis that the post-tilt bias is related to a shift in the internal estimate of direction of gravity and thus its re-adaptation when returning back upright and that it is at least to a large extent paradigm independent. This, however, does not exclude additional adaptational effects of peripheral (proprioceptive) sensory input. Previously, changes in SVV and ocular torsion during and immediately after prolonged static head-roll have been compared for head-on-trunk roll (roll-angle = 20°), demonstrating that changes in verticality perception and ocular torsion were not correlated [14]. Thus, these authors proposed an adaptation in neck proprioceptors to explain the drift observed during and after prolonged static roll-tilt. This paradigm, however, is clearly distinct from our experimental setup, with the head and trunk being aligned. Thus, adaptation of neck proprioceptors in our study seem unlikely.

4.3 Lacking effect of rotatory optokinetic stimuli on the post-tilt bias

In contrast to our initial prediction, we did not see a significant effect of the rotatory optokinetic stimulus on the size of the post-tilt bias in the SVV (or the SHV) paradigm. Thus, while such visual stimuli are very powerful in shifting the SVV when presented during the alignment task in both upright [4] and in roll-tilted positions [27], they seem to have to no prolonged effect lasting beyond the duration of stimulus presentation. Furthermore, as previously shown by our group, the effect of rotatory optokinetic stimuli on verticality perception when presented at the time of line adjustments is paradigm-dependent, being much larger for the SVV than for the SHV [5]. Thus, this further confirms our hypothesis, that rotatory optokinetic stimuli mainly affect vision-dependent mechanisms, while the internal estimate of direction of gravity is relatively unchanged.

4.4 Limitations

Our study has several limitations, including a moderate sample size of participants, a single run for each trial condition and active line/bar adjustments, requiring controlling for hysteresis in the SHV paradigm. Thus, inter-individual variability of the characteristics of the post-tilt bias including its initial amplitude and decay Tc was considerable for both paradigms. Furthermore, the previously described intra- and inter-individual variability in drift size and direction for repetitive measurements of perceived vertical further adds to the variability in the data. Noteworthy, in single subjects the drift size and drift direction seem to change little over time when repeating the same paradigm [27], suggesting that correcting for individual baseline drift in subsequent recordings is a reasonable approach.

The difference in average trial time in the SVV and the SHV paradigm (2.9±0.4 sec vs. 2.0±0.9 sec) needs to be put into relation with paradigm-specific differences, as the time for searching the bar in darkness in the SHV paradigm was not included in the total trial time.

While our experimental data supports the hypothesis that prolonged static roll-tilt resulted in a shift of the internal representation of direction of gravity, rather than being a paradigm-specific effect, this needs to be further elaborated including the implementation of model simulations for confirmation and studies addressing the individual differences in the direction of the post-tilt bias.

5 Conclusions

With a significant post-tilt bias being present in the majority of trials in both vision-dependent and vision-independent paradigms, this supports the hypothesis that prolonged static roll-tilt results in a mostly paradigm-independent shift of the internal estimate of direction of gravity. Most likely, this post-tilt bias reflects the brain’s strategy to adjust to an altered prolonged whole-body orientation relative to gravity. This bias outlasts the exposure to the stimulus (i.e., static whole-body roll-tilt) with a decay time constant in the range of one to two minutes that is fairly constant amongst paradigms. In contrast, the lacking effect of rotatory optokinetic stimuli presented during the adaptation period on the subsequent post-tilt bias speaks against a modulatory effect on the internal estimate of the direction of gravity by such a visual stimulus. This is also further supported by previous observations reporting an impact of optokinetic rotatory stimuli on simultaneous SVV adjustments but not on SHV adjustments [5].

Funding

The Betty and David Koetser Foundation for Brain Research, Zurich, Switzerland; the Center of Integrative Human Physiology, University of Zurich, Switzerland; Bonizzi-Theler-Foundation, Zurich, Switzerland; Dr. Dabbous Foundation, University of Zurich, Switzerland.

Conflict of interest

A.W., C.J.B., and A.A.T. declare no conflict of interest.

Author contributions

A.W.: collection, analysis and interpretation of data, drafting and revising the article critically for important intellectual content

C.J.B.: assisted in the design of the experiments, analysis and interpretation of data, revising the article critically for important intellectual content

A.A.T.: conception and design of the experiments, collection, analysis and interpretation of data, revising the article critically for important intellectual content

All authors have approved the final version of the manuscript, all persons designated as authors qualify for authorship and all those who qualify for authorship are listed.

Footnotes

Acknowledgments

The authors thank Marco Penner for technical assistance.

References

Brainard

D.H.

, The Psychophysics Toolbox, Spat Vis 10 (1997), 433–436.

Cullen

K.E.

and Taube

J.S.

, Our sense of direction: progress, controversies and challenges, Nat Neurosci 20 (2017), 1465–1473.

Day

R.H.

and Wade

N.J.

, Visual spatial aftereffect from prolonged head-tilt, Science 154 (1966), 1201–1202.

De Vrijer

, Medendorp

W.P.

and Van Gisbergen

J.A.

, Shared computational mechanism for tilt compensation accounts for biased verticality percepts in motion and pattern vision, J Neurophysiol 99 (2008), 915–930.

Diamond

S.G.

, Markham

C.H.

and Furuya

, Binocular counterrolling during sustained body tilt in normal humans and in a patient with unilateral vestibular nerve section, Ann Otol Rhinol Laryngol 91 (1982), 225–229.

Dichgans

, Held

, Young

L.R.

and Brandt

, Moving visual scenes influence the apparent direction of gravity, Science 178 (1972), 1217–1219.

Dockheer

K.M.

, Bockisch

C.J.

and Tarnutzer

A.A.

, Effects of Optokinetic Stimulation on Verticality Perception Are Much Larger for Vision-Based Paradigms Than for Vision-Independent Paradigms, Front Neurol 9 (2018), 323.

Eggert

, Der Einfluss orientierter Texturen auf die subjektive Vertikale und seine systemtheoretische Analyse, 1998.

Gentaz

, Baud-Bovy

and Luyat

, The haptic perception of spatial orientations, Exp Brain Res 187 (2008), 331–348.

10.

Jaggi-Schwarz

and Hess

B.J.

, Influence of dynamic tilts on the perception of earth-vertical, Exp Brain Res 149 (2003), 340–350.

11.

Kheradmand

and Winnick

, Perception of Upright: Multisensory Convergence and the Role of Temporo-Parietal Cortex, Front Neurol 8 (2017), 552.

12.

Kording

K.P.

and Wolpert

D.M.

, Bayesian integration in sensorimotor learning, Nature 427 (2004), 244–247.

13.

Laurens

and Droulez

, Bayesian processing of vestibular information, Biol Cybern 96 (2007), 405.

14.

Lechner-Steinleitner

, Interaction of labyrinthine and somatoreceptor inputs as determinants of the subjective vertical, Psychol Res 40 (1978), 65–76.

15.

Lewis

R.F.

, Priesol

A.J.

, Nicoucar

, Lim

and Merfeld

D.M.

, Dynamic tilt thresholds are reduced in vestibular migraine, J Vestib Res 21 (2011), 323–330.

16.

MacNeilage

P.R.

, Banks

M.S.

, Berger

D.R.

and Bulthoff

H.H.

, A Bayesian model of the disambiguation of gravitoinertial force by visual cues, Exp Brain Res 179 (2007), 263–290.

17.

Otero-Millan

and Kheradmand

, Upright Perception and Ocular Torsion Change Independently during Head Tilt, Front Hum Neurosci 10 (2016), 573.

18.

Pavlou

, Wijnberg

, Faldon

M.E.

and Bronstein

A.M.

, Effect of semicircular canal stimulation on the perception of the visual vertical, J Neurophysiol 90 (2003), 622–630.

19.

Pelli

D.G.

, The VideoToolbox software for visual psychophysics: Transforming numbers into movies, Spat Vis 10 (1997), 437–442.

20.

Schoene

and Udo de Haes

, Perception of gravity-vertical as a function of head and trunk position, Zeitschrift für vergleichende Physiologie 60 (1968), 440–444.

21.

Schuler

J.R.

, Bockisch

C.J.

, Straumann

and Tarnutzer

A.A.

, Precision and accuracy of the subjective haptic vertical in the roll plane, BMC Neurosci 11 (2010), 83.

22.

Shimazu

and Precht

, Tonic and kinetic responses of cat’s vestibular neurons to horizontal angular acceleration, J Neurophysiol 28 (1965), 991–1013.

23.

Tarnutzer

A.A.

, Bertolini

, Bockisch

C.J.

, Straumann

and Marti

, Modulation of internal estimates of gravity during and after prolonged roll-tilts, PLoS One 8 (2013), e78079.

24.

Tarnutzer

A.A.

, Bockisch

, Straumann

and Olasagasti

, Gravity dependence of subjective visual vertical variability, J Neurophysiol 102 (2009), 1657–1671.

25.

Tarnutzer

A.A.

, Bockisch

C.J.

and Straumann

, Head roll dependent variability of subjective visual vertical and ocular counterroll, Exp Brain Res 195 (2009), 621–626.

26.

Tarnutzer

A.A.

, Fernando

D.P.

, Kheradmand

, Lasker

A.G.

and Zee

D.S.

, Temporal constancy of perceived direction of gravity assessed by visual line adjustments, J Vestib Res 22 (2012), 41–54.

27.

Tarnutzer

A.A.

, Fernando

D.P.

, Lasker

A.G.

and Zee

D.S.

, How stable is perceived direction of gravity over extended periods in darkness? Exp Brain Res 222 (2012), 427–436.

28.

Tarnutzer

A.A.

, Schuler

J.R.

, Bockisch

C.J.

and Straumann

, Hysteresis of haptic vertical and straight ahead in healthy human subjects, BMC Neurosci 13 (2012), 114.

29.

Udo de Haes

H.A.

, Stability of apparent vertical and ocular countertorsion as a function of lateral tilt, Perception and Psychophysics 8 (1970), 137–142.

30.

Wade

N.J.

, Effect of prolonged tilt on visual orientation, Q J Exp Psychol 22 (1970), 423–439.

31.

Ward

B.K.

, Bockisch

C.J.

, Caramia

, Bertolini

and Tarnutzer

A.A.

, Gravity dependence of the effect of optokinetic stimulation on the subjective visual vertical, J Neurophysiol 117 (2017), 1948–1958.