The relative role of visual and non-visual cues in determining the perceived direction of "up": experiments in parabolic flight.

In order to measure the perceived direction of "up", subjects judged the three-dimensional shape of disks shaded to be compatible with illumination from particular directions. By finding which shaded disk appeared most convex, we were able to infer the perceived direction of illumination. This provides an indirect measure of the subject's perception of the direction of "up". The different cues contributing to this percept were separated by varying the orientation of the subject and the orientation of the visual background relative to gravity. We also measured the effect of decreasing or increasing gravity by making these shape judgements throughout all the phases of parabolic flight (0 g, 2 g and 1 g during level flight). The perceived up direction was modeled by a simple vector sum of "up" defined by vision, the body and gravity. In this model, the weighting of the visual cue became negligible under microgravity and hypergravity conditions.

Relative shear disparities and the perception of surface inclination.

A dichoptic display in which the images are cyclorotated in opposite directions does not appear inclined. This suggests that perceived inclination depends on the difference between horizontal-shear and vertical-shear disparity. Large random-dot stereoscopic displays were presented with various types of shear disparity. Perceived inclination was the same magnitude for horizontal and vertical shear disparities. Opposed horizontal and vertical shear produced greater inclination than a single-axis shear. Same-sign vertical and horizontal shear (rotation) produced no inclination. These results support the relative-shear hypothesis. Cyclovergence was measured and was insufficient to account for these effects. We conclude that perceived inclination depends on the difference between horizontal- and vertical-shear disparities. Perceived inclination was not based on vertical disparity within small displays or within large displays with a zero-disparity surround. Relative-shear disparities are therefore extracted globally rather than locally.

Human optokinetic nystagmus in response to moving binocularly disparate stimuli.

Physiological and behavioral evidence shows that the directionally preponderant subcortical control of optokinetic nystagmus (OKN) in lower mammals is supplemented in higher mammals by bidirectional cortical control. It is hypothesized that this cortical control allows higher mammals to cope with the parallactic movement of the scene produced by linear motion of the body. In particular, it is hypothesized that a coupling between OKN and stereopsis allows higher mammals to stabilize the images of objects within the plane of fixation while ignoring motion signals from objects at other distances. According to this hypothesis the gain of the slow phase of OKN should be highest for binocularly fused moving stimuli and attenuated for binocularly disparate displays. The results of Experiment 1 confirmed this prediction although the effects of accommodation were not ruled out completely. In Experiment 2 a display moving in one direction was presented across the central retina at the same time as one moving in the opposite direction was presented in the upper and lower periphery. It was found that subjects do not show OKN in the direction of the peripheral display unless it is binocularly fused and the central display is disparate. In Experiment 3 a stationary display of dots was superimposed on a moving display. It was found that OKN is not inhibited by the stationary display when it has a horizontal disparity and the moving display is fused. Experiment 4 found that horizontal OKN is disrupted by the sudden introduction of a vertical disparity in the stimulus. Since accommodative state was kept constant in the last three experiments, the data show that binocular disparities can help a person to stabilize selectively the image of one moving display while ignoring conflicting motion signals from another display.

Effects of disparity-perspective cue conflict on depth contrast.

The role of disparity-perspective cue conflict in depth contrast was examined. A central square and a surrounding frame were observed in a stereoscope. Five conditions were compared: (1) only disparity was introduced into either the centre or surround stimulus, (2) only perspective was introduced into the centre or surround, (3) concordant perspective and disparity were introduced into the centre or surround, (4) disparity was introduced into one stimulus and perspective into the other, and (5) only the centre stimulus was presented with horizontal shear disparity and perspective manipulated independently. The results show that individual differences in depth contrast were related to individual differences in the weighting of disparity and perspective in the single-stimulus conditions. We conclude that conflict between disparity and perspective contributes to depth contrast. However, significant depth contrast occurred when there was no disparity-perspective cue conflict, indicating that this cue conflict is not the sole mechanism producing depth contrast.

Optokinetic torsion: dynamics and relation to circularvection.

Continuous records of optokinetic torsion to sinusoidal inputs were obtained using the electromagnetic scleral search-coil technique. We measured the gain and phase lag of optokinetic torsion in response to a spherical visual display rotating steadily at various angular velocities and sinusoidally at frequencies from 0.2 to 2.0 Hz and at amplitudes from 10 to 80 deg. Gain (peak slow-phase eye velocity over stimulus angular velocity) of up to 0.12 were obtained with stimulus frequencies of 0.2 Hz and declined to an average value of about 0.02 at a frequency of 2.0 Hz. Phase lag was virtually zero at a frequency of 0.2 Hz and increased to over 80 deg at 2.0 Hz. The records from the sinusoidal stimuli show very few quick phases. With increasing stimulus amplitudes, the amplitude of the response increased but its gain declined. We found no evidence of torsional after-nystagmus nor any relation between the torsional response and reports of vection or sensation of body tilt induced by the rotating display. Torsional optokinetic nystagmus is most suited to compensate for low-amplitude, low-frequency stimulus rotation and normally supplements torsion induced by head tilt.

Stereopsis with persisting and dynamic textures.

We measured the percept of changing depth from changing disparity in stereograms composed of random-dot textures that were either persistent or dynamically changed on every frame (a dynamic random-dot stereogram). Disparity was changed between frames to depict a surface undergoing smooth temporal changes in simulated slant. Matched depth was greater with dynamic random-dot stereograms than with persistent random-dot stereograms. These results confirm and extend earlier observations at depth threshold. We posit an explanation based on cue conflict between stereopsis and monocular depth cues.

Spatial properties of shear disparity processing.

We investigated whether vertical-shear disparity was extracted from the whole visual field or from a more local area and how global estimates of vertical disparity are derived. We also investigated the role of cyclovergence in processing shear disparity. Random-dot stereoscopic displays in various configurations were presented with horizontal-shear disparity, vertical-shear disparity or same-sign horizontal- and vertical-shear (rotation) disparity. Vertical-shear disparity introduced into only the right half of a 60 deg-wide display produced perceived inclination of the whole display when the center of shear was on the fovea, but did not produce inclination, either of the whole display or of a local area when the centre of shear was in an eccentric retinal position. A display containing dots with vertical-shear disparity mixed with dots with zero-disparity produced one inclined surface. Horizontal-shear disparity always produced inclination confined to the local area of disparity. Rotation disparity produced no inclination when introduced into the whole display, but when introduced with zero-disparity dots. It produced an inclined plane distinct from the plane defined by the zero-disparity dots. These results could be attributed to cyclovergence, which we therefore eliminated in our last experiment. We conclude that the perception of surface inclination is based on the difference between local horizontal-shear disparity and global vertical-shear disparity averaged over the whole visual field.

Temporal dependencies in resolving monocular and binocular cue conflict in slant perception.

Observers viewed large dichoptic patterns undergoing smooth temporal modulations or step changes in simulated slant or inclination under various conditions of disparity-perspective cue conflict and concordance. After presentation of each test surface, subjects adjusted a comparison surface to match the perceived slant or inclination of the test surface. Addition of conflicting perspective to disparity affected slant and inclination perception more for brief than for long presentations. Perspective had more influence for smooth temporal changes than for step changes in slant or inclination and for surfaces presented in isolation rather than with a zero disparity frame. These results indicate that conflicting perspective information plays a dominant role in determining the temporal properties of perceived slant and inclination.

Relative size disparities and the perception of surface slant.

Perceived slant produced by size disparities in random-dot displays was measured by tactile matching. For a 60 deg surface, slant produced by vertical-size disparity (the induced effect) was opposite to that produced by horizontal-size disparity. Overall-size disparity produced a little slant. With small displays, effects of horizontal and vertical disparities were reduced but not those of overall disparity. A zero-disparity surround increased effects of horizontal and overall disparities but reduced the induced effect. A mixture of horizontally disparate and zero-disparity dots produced two slanted surfaces. Vertically disparate and zero-disparity dots produced one slanted surface. Abutting opposite horizontal disparities produced surfaces with a sharp boundary. Abutting vertical disparities produced surfaces with a gradual boundary. Perceived slant depends on the difference between horizontal-size disparity detected locally and mean vertical-size disparity over a relatively large area.

Spatial limitation of vertical-size disparity processing.

We investigated the upper limit of horizontal spatial modulation of vertical-size disparity in a textured surface for the perception of depth. In Experiment 1 subjects matched the appearance of a surface with modulated horizontal-size disparity to that of a surface with modulated vertical-size disparity. In Experiment 2 we determined the threshold amplitude of modulation of vertical-size disparity required for the perception of depth as a function of the spatial frequency of disparity modulation. The results indicate that sensations of depth are not elicited by modulations of vertical-size disparity of any amplitude at spatial frequencies higher than about 0.04 c/deg. We conclude that vertical disparities are averaged within about 20 deg-wide areas and suggest that this global measurement is used to scale local horizontal disparities for the perception of surface slant.

The efficiency of the central and peripheral retina in driving human optokinetic nystagmus.

Previous experiments to decide whether the gain of optokinetic nystagmus (OKN) is increased or decreased by occlusion of the central retina involved the use of stationary edges on the occluder and unmatched contrasts. With these factors controlled, it was confirmed that OKN gain is severely reduced by occlusion of the central retina but only at stimulus velocities above about 30 degrees/sec. The gain of horizontal OKN was found not to increase with increasing width of the display if the lateral edges are blurred. The high gain of centrally driven OKN may be related to the ability of higher mammals to stabilize the images of objects at a given distance in a complex parallactic visual field.

Optokinetic nystagmus: the effects of stationary edges, alone and in combination with central occlusion.

In Experiment 1 we investigated the independent and combined effects of horizontal OKN of stationary edges and occlusion of the central retina. For a display 60 degrees wide moving at 30 degrees/sec a symmetrically placed pair of vertical nonoccluding bars suppressed OKN when near the center of the display but had no effect when 30 degrees apart. A 7 degrees-high 60 degrees-wide central occluder reduced OKN gain by 37%. However, a central occluder with edges only 30 degrees wide abolished OKN. In Experiment 2 this interaction between central occlusion and stationary edges was confirmed with a wider display over a range of stimulus velocities and configurations. A functional explanation of this interaction is presented.

Depth selectivity of vertical fusional mechanisms.

We measured the ability to fuse dichoptic images of a horizontal line alone or in the presence of a textured background with different vertical disparity. Nonius-line measurements of vertical vergence were also obtained. Diplopia thresholds and vertical vergence gains were much higher in response to an isolated vertically disparate line than to one with a zero vertical-disparity background. The effect of the background was maximum when it was coplanar with the target and decreased with increasing relative horizontal disparity. We conclude that vertical disparities are integrated over a restricted range of horizontal disparities to drive vertical vergence.

Effects of horizontal and vertical additive disparity noise on stereoscopic corrugation detection.

Stereoscopic corrugation detection in the presence of horizontal- and vertical- additive disparity noise was examined using a signal detection paradigm. Random-dot stereograms either represented a 3-D square-wave surface with various amounts of Gaussian-distributed additive disparity noise or had the same disparity values randomly redistributed. Stereoscopic detection of 2 arcmin peak amplitude corrugations was found to tolerate significantly greater amplitudes of vertical-disparity noise than horizontal-disparity noise--irrespective of whether the corrugations were horizontally or vertically oriented. However, this directional difference in tolerance to disparity noise was found to reverse when the corrugation and noise amplitudes were increased (so as to produce equivalent signal-to-noise ratios). These results suggest that horizontal- and vertical-disparity noise pose different problems for dot-matching and post-matching surface reconstruction as corrugation and noise amplitudes increase.

Induced rotary motion and ocular torsion.

When a large patterned annulus rotates around a stationary sectored disc the latter appears to rotate in the opposite direction. Such induced rotary motion was examined with central discs subtending 5, 20 and 40 deg at the eye, with the surround filling the remainder of the visual field. The annular surround or the central disc could be oscillated sinusoidally around the fixation point through 20 deg at 0.2 Hz. In each case, subjects estimated the angles through which the moving and stationary parts of the display appeared to rotate on one half-cycle. Subjects also estimated the angle of rotation of an oscillating display that filled the visual field. Induced rotation of the centre was around 100% of the inducing amplitude for all disc sizes, but there was no induced motion of the surround when the centre rotated. Ocular torsion was measured under the same conditions, using the scleral search-coil technique. The amplitude of ocular torsion was a function of the size of the stationary or rotating field. Thus, variations in stimulus conditions affected induced rotary motion and ocular torsion in different ways. The implications of the results for theories of induced motion in terms of underregistered eye movements are discussed.

Circularvection about earth-horizontal axes in bilateral labyrinthine-defective subjects.

A stationary subject surrounded by a visual display rotating about an earth-horizontal axis typically experiences a sensation of continuous self-rotation (vection) coupled with a paradoxical sensation of a limited degree of body tilt, both opposite to the direction of the stimulus. The sensation of limited body tilt has been attributed to conflict between visually-induced vection, and otolithic and somatosensory graviceptive information which indicates that the body has not moved. We investigated circularvection and illusory body tilt about the horizontal axis in the pitch and roll planes in bilateral labyrinthine-defective (L-D) subjects. Results demonstrated that the bilateral group experienced complete unambiguous self-rotation through an upside-down orientation. The relative contributions of the otolithic and somatosensory graviceptors to visuall-induced tilt is discussed.

Interactions within and between the spatial senses.

This paper reviews five types of interaction between sources of spatial information within and between sense organs; 1) nested, 2) opponent, 3) comparison, 4) covariation, and 5) multicue interactions. Efference copy is treated as a type of sensory input. Examples of each type of interaction are provided, with an emphasis on visual-vestibular interactions. In the first type of interaction, inputs from nested sensory systems are summed like vectors. For instance, the 3-D vector sum of inputs from the joints and muscle spindles of the arm allows one to judge the position of the hand. In the second type, inputs from opponent systems are combined to form a signed difference signal with respect to a norm. For instance, the push-pull linkage between the vestibular organs on the two sides of the head provides the signal for head rotation. The third type involves comparisons based on the detection of differences between stimuli presented to different regions of the same sense organ or to distinct sense organs. The fourth type involves the extraction of products or ratios between stimuli used in the detection of invariant high-level features. For instance, the linear size of an object can be derived from the constant product of the distance of the object and the size of its image. Similar systems are used to scale the response to one stimulus feature with respect to a second feature. For instance, vestibular inputs evoking eye nystagmus are scaled by viewing distance. Judgments based on all of the above mechanisms are relational, meaning that they require information from several sources. The fifth type involves multicue systems in which alternative cues are available for the same judgment. The cues are sometimes combined as a weighted mean. For instance, the direction of an object is derived from the mean position of the images in the two eyes, or a judgment of the rotation of the body may be based on combined inputs from the vestibular system and from visual motion. For distinct types of cue, averaging is less common than cue dominance, dissociation, or cue reinterpretation.

Human gaze instability during brief exposure to reduced gravity.

The stability of gaze in three dimensions (horizontal, vertical and torsional) was investigated with the electromagnetic scleral search-coil technique during the microgravity phase of parabolic flights under two visual fixation conditions: fixation on a real target and on an imagined target. Subjects were secured upright with the head immobilized by a dental bite. There were torsional eye movements in response to the imposition of reduced gravitoinertial forces under both visual fixation conditions. The pattern of these movements was consistent with our previous findings on six other subjects. No significant horizontal eye movements were observed in either fixation condition. Under the condition of fixation on an imagined target, direction-specific vertical nystagmus was observed with slow phase directed upwards during transition from hypergravity to microgravity. The slow phase was directed downwards during transition from microgravity to hypergravity, although the nystagmus was of lower frequency and the magnitude of the slow phase velocity appeared to be smaller than during transition from hypergravity to microgravity. The vertical eye movements could be attributed to a change of otolithic stimulation along the subject's z axis since the rate of pitch rotation of the aircraft during parabolic flight was too slow to produce an effective canal input. These kinds of reflex eye movements could degrade vision during manoeuvres or turbulence in flight.

Post-rotatory nystagmus and turning sensations after active and passive turning.

We measured post-rotatory nystagmus and sensations of body rotation in standing subjects brought to rest in the dark after 3 minutes of each of the following conditions: 1) passive turning about the mid-body axis, involving only vestibular stimulation, 2) active turning about the mid-body axis, involving both vestibular stimulation and motor-proprioceptive activity in the legs, and 3) stepping round while remaining facing in the same direction on the center of a rotating platform with the head held in a stationary holder (apparent turning), involving only motor-proprioceptive activity. The same acceleration-velocity profile was used in all conditions. Post-rotatory nystagmus (slow phase) occurred in the same direction to passive body turning and was reduced in velocity after active body turning. After apparent turning, nystagmus was in the opposite direction as attempted body turning. Our theoretical analysis suggests that nystagmus after active turning should conform to the mean of the responses after passive and apparent turning rather than to their sum. The results conform more closely to the mean than to the sum, but with greater weight given to vestibular inputs than to motor-proprioceptive inputs. Post-rotatory sensations of self-rotation were in the expected opposite direction after passive turning and were lower in magnitude after active turning. After apparent turning, sensations of self-rotation were in the same direction as those after attempted turning--an effect known as the antisomatogyral illusion.

Relative role of visual and non-visual cues in determining the direction of "up": experiments in the York tilted room facility.

Perceiving a direction as "up" is fundamental to human performance and perception. Astronauts in microgravity frequently experience reorientation illusions in which they, or their world, appear to flip and 'up' becomes arbitrarily redefined. This paper assesses the relative importance of visual cues in determining the perceived up direction. In the absence of information about the origin of illumination, people interpret surface structure by assuming that the direction of illumination is from above. Here we exploit this phenomenon to measure the influence of head and body orientation, gravity and visual cues on the perceived up direction. Fifteen subjects judged the shape of shaded circles presented in various orientations. The circles were shaded in such a way that when the shading was compatible with light coming from above, the circle appeared as a convex hemisphere. Therefore, by finding which shaded circle appeared most convex, we can deduce the direction regarded as "up". The different cues contributing to this percept were separated by varying both the orientation of the subject and the surrounding room relative to gravity. The relative significance of each cue may be of use in spacecraft interior design to help reduce the incidence of visual reorientation illusions.