Changes to Control of Adaptive Gait in Individuals with Long-standing Reduced Stereoacuity

Abstract

Purpose.: Gait during obstacle negotiation is adapted in visually normal subjects whose vision is temporarily and unilaterally blurred or occluded. This study was conducted to examine whether gait parameters in individuals with long-standing deficient stereopsis are similarly adapted.

Methods.: Twelve visually normal subjects and 16 individuals with deficient stereopsis due to amblyopia and/or its associated conditions negotiated floor-based obstacles of different heights (7–22 cm). Trials were conducted during binocular viewing and monocular occlusion. Analyses focused on foot placement before the obstacle and toe clearance over it.

Results.: Across all viewing conditions, there were significant group-by-obstacle height interactions for toe clearance (P < 0.001), walking velocity (P = 0.003), and penultimate step length (P = 0.022). Toe clearance decreased (∼0.7cm) with increasing obstacle height in visually normal subjects, but it increased (∼1.5 cm) with increasing obstacle height in the stereo-deficient group. Walking velocity and penultimate step length decreased with increasing obstacle height in both groups, but the reduction was more pronounced in stereo-deficient individuals. Post hoc analyses indicated group differences in toe clearance and penultimate step length when negotiating the highest obstacle (P < 0.05).

Conclusions.: Occlusion of either eye caused significant and similar gait changes in both groups, suggesting that in stereo-deficient individuals, as in visually normal subjects, both eyes contribute usefully to the execution of adaptive gait. Under monocular and binocular viewing, obstacle-crossing performance in stereo-deficient individuals was more cautious when compared with that of visually normal subjects, but this difference became evident only when the subjects were negotiating higher obstacles; suggesting that such individuals may be at greater risk of tripping or falling during everyday locomotion.

Movement control is a complex process that involves modulation of motor output according to sensory information provided by the visual, vestibular, and proprioceptive systems. ¹ Since the visual system provides information not only about current surroundings but also about future movement and obstacle-avoidance challenges, it is thought to be the predominant sense used in planning movements in, and through, the environment. ² Adult humans have two eyes that are horizontally separated by approximately 6 cm, but our percept is cyclopean and arises from a combination of the images that fall on the retina of the right and left eyes. In a properly functioning visual system, the integration of visual information from the eyes also gives rise to stereopsis, which conveys information about the relative distance of objects from the observer. Stereopsis is derived directly from the disparate retinal images formed in the right and left eyes, and consequently this form of depth information arises only as a result of binocular cooperation. Indeed, stereopsis is often considered as sitting at the top of the food chain of vision ³ (for recent review, see Ref. 4) and as a barometer of binocular cooperation. ⁵ It is thus natural to enquire how reduced or absent stereopsis affects everyday movement control of the entire body (e.g., negotiating a ground-based obstacle when walking) or of parts of the body (e.g., reaching for a cup). 

In the investigation of the importance of stereopsis in the control of everyday mobility, several studies have been undertaken to examine the consequences, for obstacle-crossing performance, of reducing or eliminating stereopsis in visually normal individuals by covering one eye, ^6,7 or by degrading vision through unilateral ^7,8 or bilateral defocus. ^9,10 Vale et al. ⁸ demonstrated that monocular defocus leads to gait changes that are similar to those found when one eye is occluded. However, in their participants, vision was unilaterally blurred only for the duration of the experiment, and hence the impact of chronic monocular blur on everyday mobility (as in monovision correction) is currently unknown. These visual manipulations are easy to implement, but this experimental approach suffers from at least two disadvantages. The first is that stereopsis is not the only visual attribute affected when one eye is covered or when one or both eyes are blurred. A second and more important disadvantage is that visually normal subjects with blurred or uniocularly occluded vision may not behave in the same way as those who have endured degraded or absent stereopsis from early childhood. Thus, an arguably better approach to studying the effects of degraded or absent stereopsis on everyday mobility is to compare adaptive gait in visually normal subjects and individuals displaying reduced or absent stereopsis that has arisen naturally and is of long-standing duration. One of the few studies in which this approach was used was an investigation of performance on a mobility task before and after second-eye cataract surgery. ¹¹ In this study it was reported that surgery led to improved stereopsis and to better mobility performance. In the present study, we examined mobility in individuals with long-standing stereo-deficiency due to amblyopia and/or the conditions thought to cause it. 

Amblyopia is a developmental disorder of vision that may be defined as reduced visual acuity in one eye or, infrequently, in both eyes that is not due to refractive error, disease, or structural defects of the eye or visual system. ¹² Depending on the population studied and diagnostic criteria applied ¹³ the prevalence of amblyopia is ∼2% to 2.5%, ¹² and it represents the most common cause of monocular visual impairment among children and young adults. ¹⁴ As well, having a reduction in best corrected visual acuity, individuals with amblyopia exhibit abnormal binocularity, often with grossly reduced or absent stereopsis. ¹⁵ Other naturally occurring causes of stereo-deficiency are strabismus (ocular misalignment) and anisometropia (a significant difference in refractive error) and, if present in early childhood, these conditions are believed to represent the cause in nearly all cases of human amblyopia. ^13,16  

Given the relatively high prevalence of amblyopia and its associated conditions, ^12,13,16 it is surprising that the functional consequences relating to movement control in affected individuals have only recently begun to be studied. Although the effects of naturally occurring, reduced stereopsis on prehension ^17,18 and other hand–eye co-ordination tasks ¹⁹ have recently been examined, no studies have been conducted to investigate how gait may be affected. By measuring foot positioning and toe clearance during obstacle negotiation, we sought in the present study to determine whether individuals with long-standing stereo-deficiency adapt their gait during everyday locomotion. By comparing performance under binocular viewing conditions with that exhibited when the nondominant eye was occluded, we also sought to establish whether in the control of gait, as in visually normal subjects, two eyes are better than one in these individuals. 

Methods

Visual Assessment

All participants underwent subjective refraction and binocular vision assessment that included cover test, ocular motility, and stereo threshold (using the near Frisby stereotest, a stereotest that relies on real-depth rather than apparent depth ²⁰ ) assessments (Table 1). Ocular dominance was determined in all participants with the Kay-pictures Dominant Eye Test (www.kaypictures.co.uk/dominant.html) and was classified as the eye that was used for sighting on at least two of the three presentations. Contrast sensitivity (CS) was measured by using the Pelli-Robson chart ²¹ (200 cd/m⁻²) and a per-letter scoring system. ²² Two versions of the chart were used to minimize learning, and the testing distance was 2 m. Visual acuity was measured with a computer-based system (Test Chart 2000; Thomson Software Solutions, Herts UK) at a testing distance of 3 m and a per-letter scoring system. Visual acuity (VA) and CS measurements (Table 1) were taken in binocular and monocular conditions with the participant's habitual refractive correction in place. 

Table 1.

 

Clinical Details Relating to Stereo-deficient Participants, of Whom Nine Were Amblyopic and Seven Were Not

Subject	Age	Dominant Eye	Refraction		Visual Acuity			Strab	Oculomotor Status	Stereo-acuity	Contrast Sensitivity			Height (cm)
			LE	RE	Hab	D	ND				Hab	D	ND
Amblyopes
AH	23	Left	+0.25/+0.25 × 90	+1.50/+0.25 × 175	0.00	+0.02	+0.24	Y	8 RSOT	Negative	1.70	1.65	1.60	163
CC	41	Right	−1.25/+1.25 × 90	−3.00/+0.25 × 170	0.00	0.00	+1.50	Y	8 LXOT	Negative	1.85	1.70	0.00	162
CM	26	Right	+4.00/+1.75 × 20	+4.00/+2.00 × 158	0.00	0.00	+0.50	Y	10 LSOT	Negative	1.70	1.80	1.65	145
GH	54	Right	+1.00/+0.25 × 180	+4.50/+0.25 × 170	−0.10	−0.10	+0.58	Y	14 LSOT	Negative	1.80	1.80	1.20	160
JC	30	Right	+1.25 DS	+3.75/+2.50 × 145	−0.10	−0.10	+0.86	N	6 XOP	Negative	1.90	1.80	1.45	161
JT	46	Right	+0.75/+0.25 × 150	+2.75/+1.00 × 130	−0.12	−0.10	+0.40	N	4 XOP	300 sec/arc	1.65	1.65	1.05	180
LA	21	Right	+2.75/+1.00 × 100	+4.75/+1.25 × 95	−0.10	−0.10	+0.66	N	2 SOP	85 sec/arc	1.65	1.65	1.65	171
RC	46	Left	−5.00/+5.50 × 108	0.00/+0.25 × 90	0.00	0.00	+0.80	N	2 XOP	85 sec/arc	1.80	1.70	1.35	178
UM	36	Right	−2.50/+0.50 × 165	−3.00 DS	−0.04	0.00	+0.22	Y	6 LXOT	120 sec/arc	1.75	1.70	1.60	154
Nonamblyopic deficient stereopsis
CN	23	Right	+1.00/+0.50 × 82	+4.25/+0.50 × 44	−0.04	0.00	0.00	N	2 XOP	600 sec/arc	1.85	1.75	1.60	178
FD	29	Right	+0.75/+0.50 × 125	+0.75 DS	−0.12	−0.10	0.00	Y	8 LSOT	Negative	1.80	1.80	1.45	172
HK	41	Right	−2.50/+0.25 × 136	−2.50/+0.25 × 141	−0.16	−0.10	0.00	Y	18 LSOT	Negative	1.80	1.70	1.75	156
HR	21	Left	+5.00/−0.25 × 26	+4.75/−0.25 × 132	−0.14	−0.10	−0.10	Y	12 ALT SOT	Negative	1.85	1.90	1.70	162
KB	21	Left	−2.25/+0.25 × 90	−1.50/+0.75 × 100	0.00	0.00	+0.10	Y	14 R/ALT XOT	Negative	1.80	1.70	1.65	176
PF	38	Left	+0.25 DS	−1.00/+0.75 × 130	−0.14	−0.10	0.00	Y	20 R/ALT XOT	Negative	1.70	1.65	1.65	168
SO	58	Left	+3.75/+0.50 × 175	+0.75/+0.75 × 175	−0.20	−0.20	0.00	N	2 XOP	170 sec/arc	1.75	1.65	1.65	163
Visually normal
Mean ± SD	31 ± 8.59				−0.07 ± 0.07	−0.05 ± 0.08	−0.04 ± 0.08	N		28.33 ± 10.33	1.80 ± 0.09	1.74 ± 0.07	1.73 ± 0.10	167 ± 15.30

 VA, monocular visual acuity VA (logMAR); CS contrast sensitivity (Pelli Robson) data are presented for habitual (both eyes open, Hab) viewing, dominant eye (D) viewing and non-dominant eye (ND) viewing conditions. Oculomotor status in distance viewing is presented in prism diopters, and stereoacuity is given in seconds of arc (negative indicates worse than 600 sec/arc on Frisby stereoacuity test). Mean data (±SD) for the 12 visually normal participants are also presented. SOT, esotropia; XOT, exotropia; SOP, esophoria, XOP.

Participants

A total of 28 participants took part in the study. Twelve were visually normal (mean age, 31 ± 8.6 years; three men and nine women; mean height, 169 ± 15 cm; mean weight, 63 ± 13.7 kg) and they comprised the control group against which the stereo-deficient (DS) group (n = 16; mean age, 34.6 ± 12 years; 1 man/15 women; mean height, 163 ±18 cm; mean weight, 70 ± 14 kg) was compared. Participants were recruited from the staff and student population at the University of Bradford and from the surrounding area. Informed, written consent was obtained from all participants before their participation. The study was approved by the local bioethics committee, and complied throughout with the tenets of the Declaration of Helsinki. 

Exclusion criteria included a history of neurologic, musculoskeletal, or cardiovascular disorders that could affect balance or gait, or a history of ocular disease (with the exception of strabismus in the DS group). Visually normal participants, with habitual correction if worn, had monocular VA of at least 6/6 (0.0 logMAR) in each eye and stereopsis of at least 60 sec/arc; the mean ± SD stereoacuity for the visually normal group was 28.3 ±10.3 sec/arc. Nine of the DS group had unilateral amblyopia, defined as the acuity of one eye being below 6/6 with a 2-line (or greater) acuity difference between the best corrected VA of the right and left eyes. Only four amblyopes had measurable stereoacuity and, of these, three were nonstrabismic. The remaining seven DS participants had approximately equal VA (dominant eye [D], −0.09 ± 0.07 logMAR; nondominant eye [ND] 0.00 ± 0.06 logMAR), and stereoacuity was measurable in only two of them (CN and SO, both nonstrabismic). Participants CN and SO had long-standing reduced stereopsis associated with substantial anisometropia, whereas the remaining five participants exhibited strabismus (Table 1). For the individuals with amblyopia, the median acuity for the dominant/fellow eye was −0.04 logMAR and the median acuity of the amblyopic/nondominant eye was +0.58 logMAR (Table 1). 

The visually normal and DS groups did not differ in relation to age (P = 0.18) or height (P = 0.19) but, as expected, they differed significantly in stereo threshold (P < 0.001, Table 1). When subjects viewed with the dominant eye alone, there was no significant difference between groups in relation to VA (P = 0.36) or CS (P = 0.21). Similarly, when the subjects viewed binocularly, VA (P = 0.40) and CS (P = 0.15) did not differ between the groups. 

Protocol

Three-dimensional body segment kinematic data were collected (at 100 Hz) with an eight-camera, motion-capture system (Vicon MX; Oxford Metrics, Oxford, UK) as each subject walked across the laboratory and negotiated a floor-based obstacle placed in the travel path, approximately five walking steps away from where they began walking. The laboratory was illuminated with ceiling-mounted fluorescent strip lighting. The illumination over the walking area averaged 400 lux, and the luminance of the floor was 15 cd · m⁻². Participants wore their own low-heeled shoes that they deemed comfortable for walking. Twenty-seven retroreflective markers (14-mm diameter) were attached directly onto either the skin or clothing at the following locations (as per Plug-In Gait software guidelines; Vicon Oxford Metrics): lateral malleoli, lateral aspects of each shank and thigh, lateral femoral condyles, anterior superior iliac spines, sacrum, medial and lateral aspects of the wrist, lateral humeral epicondyles, acromions, inferior tip of the sternum, jugular notch, spinous processes of the 7th cervical and 10th thoracic vertebrae, and the anterolateral and posterolateral aspects of the head. Markers (6-mm diameter) were also attached to each shoe, corresponding to the following locations: superior aspects of the second and fifth metatarsal heads and end of the second toe and to the upper edge of the obstacles, to determine its location within the laboratory coordinate system. Three different obstacle heights (7, 15, and 22 cm) were used. To encourage participants to use visual information to determine the position of the obstacle within the travel path rather than simply repeat a motor strategy to negotiate the obstacle, its fore–aft position was varied (i.e., blind to participant) from trial to trial by 0, 10, 20, or 30 cm. For the same reason, 15 randomly presented catch trials (i.e., no obstacle present) were also undertaken. The participants were instructed to walk at their customary walking pace to the opposite end of the laboratory, stepping over the obstacle, if present. Besides this instruction, no other specific instruction was provided. Head or gaze movements were not controlled or monitored, and participants were not asked to lead with any particular limb. Trials were completed with the participant's distance refractive correction (if worn) under binocular (Bin), nondominant eye (ND), and dominant eye (D) viewing conditions. A white patch, taped directly onto the participant's skin, was used for occlusion. A total of 87 trials [(3 heights × 3 vision conditions × 8 repetitions) + 15 catch trials] were conducted in a pseudorandom order for each participant. Data collection for each participant took approximately 2 hours, including rest periods. 

Data Analysis

Using the gait software (Plug-In-Gait; Vicon Oxford Metrics), we filtered the marker trajectory data with the Woltring spline-smoothing routine ²³ with the mean square error (MSE) filter option set to 10. Data were then processed to define a 3-D link–segment model of the subject. A virtual shoe tip marker representing the inferior tip of the shoe was determined by reconstructing its position relative to the markers placed on the second and fifth metatarsal heads and end of the second toe (Fig. 1). The 3-D coordinate data of the sternum, each of the foot markers (including the virtual shoe tip), and the markers placed on the obstacle were exported in ASCII format for further analysis. 

Figure 1.

 

Movement of the leading foot (shaded) as the obstacle (rectangle) is crossed. Trailing-foot (unshaded) positioning is also shown. In each trial, VTC and foot placement parameters including leading-foot distance, trailing-foot distance, penultimate step length, and crossing step length were determined. Approach/walking velocity was also calculated.

Analysis focused on foot placement parameters during the approach and toe clearance parameters over the obstacle, as shown in Figure 1. Walking velocity was also estimated by calculating the instantaneous forward velocity of the sternum marker and then calculating an average value over the duration of the penultimate step. 

Statistical Analysis

Before the statistical analyses were performed, the data set was inspected to eliminate clear outliers (>3 SD from the group mean). Most of these were errors resulting from problems with motion/kinematic tracking, for example, because body segment markers momentarily disappeared from camera view. Both subject groups (normals and DS) had a similar proportion of excluded data points (<2% of data). After these deletions, the resulting data sets were shown to be normally distributed according to the Shapiro-Wilks test (P > 0.05). 

For each dependent measure (averaged across the eight repetitions), repeated-measures ANOVAs were used to determine main and interaction effects for the following:

•  
  Group (G): two levels (visually normal subjects, DS individuals).
•  
  Visual condition (V): three levels (binocular, dominant, and nondominant viewing).
•  
  Obstacle height (H): three levels (7, 15, and 22 cm).

For all analyses the α-level for statistical significance was set at P ≤ 0.05. Post hoc analyses of significant vision or obstacle height effects were undertaken with Tukey's HSD test. Post hoc analyses of significant group-by-obstacle-height or group-by-vision interactions were undertaken with the t-test.

Our DS group was made up of individuals with and without amblyopia (Table 1). To confirm that we were justified in grouping together the results for our stereo-deficient individuals with and without amblyopia into one DS group, we used a preliminary statistical analysis to compare main outcome measures between these subgroups. This analysis indicated that there were no significant differences (all P > 0.21) for any of the outcomes measures between DS individuals with or without amblyopia. Henceforth, therefore, all the results for our DS individuals are considered together as a single group. 

Results

Although the focus of the results presented herein are on gait similarities and differences between the visually normal and DS groups, we also examined the trial-by-trial variability in gait parameters in each group, but found no significant difference between them (all P > 0.35). 

Differences in Adaptive Gait between Stereo-deficient and Visually Normal Individuals

There were no main effects of group (DS versus normal) on any of the gait parameters investigated (Table 2). However, there were significant group-by-step height interactions for vertical toe clearance (VTC; P < 0.001), walking velocity (P = 0.003), and penultimate step length (P = 0.022, Table 2). Across all viewing conditions, VTC was found to decrease by ∼0.7 cm with increasing step-height in visually normal subjects, but it increased by ∼1.5 cm with increasing step height in the DS group (Fig. 2), and post hoc analyses indicated the toe clearance was significantly different between groups when negotiating the highest obstacle (P < 0.05). (We checked that the toe rather than the heel was the closest part of the foot to the obstacle during crossing, by checking vertical heel clearance. This analysis confirmed that the heel was always higher than the toe at the point of crossing.) In both groups, gait parameters were significantly affected by monocular occlusion. Walking velocity and penultimate step length decreased with increasing step height (again, across all viewing conditions) in both groups, but the reduction was much more pronounced in the DS group (walking velocity decrease: ∼52 cm/s for visually normal subjects versus ∼105 cm/s for the DS-group; penultimate step length decrease: ∼6 cm for visually normal subjects versus ∼15 cm for the DS-group; Fig. 3). Post hoc analyses of the significant group-by-obstacle height interactions indicated penultimate step length was significantly different between groups when negotiating both the medium and high obstacles (P < 0.05). The difference between groups in walking velocity did not reach statistical significance when negotiating any of the obstacle heights, although group differences for the highest obstacle approached statistical significance (P = 0.06). There was also a significant group-by-vision-by-obstacle-height interaction (P = 0.04, Table 2), indicating that VTC in DS individuals under dominant eye viewing was no different from that in binocular viewing when the obstacle height was low. To determine whether differences found across all viewing conditions were also present when the habitual viewing condition was considered alone, we also performed post hoc analyses of the data collected for just the binocular viewing condition. This analysis indicated group differences in penultimate step length when negotiating the medium and high obstacles (P < 0.05), group differences in VTC when negotiating the medium height obstacle and a trend at the high obstacle (P = 0.079), but no group differences in walking velocity (P > 0.16). 

Table 2.

 

Gait Parameters for the Visually Normal and DS Groups

	Visually Normal			Deficient Stereopsis			ANOVA
	Bin	D	ND	Bin	D	ND	Main Effects	Interactions
VTC, mm	98 ± 43	116 ± 50	123 ± 46	121 ± 39	130 ± 38	142 ± 37	G (P = 0.24)	gv (P = 0.30)
		*Bin (P < 0.001)	Bin (P < 0.001)		Bin (P < 0.001)	Bin (P < 0.001)	V (P < 0.001)	gh (P = 0.0007)
		ND (P = 0.001)	D (P = 0.001)		ND (P = 0.001)	D (P = 0.001)	H (P < 0.001)	vh (P = 0.64)
								gvh (P = 0.04)
Leadfoot distance, mm	847 ± 159	820 ± 168	814 ± 158	783 ± 176	740 ± 199	727 ± 191	G (P = 0.16)	gv (P = 0.64)
		Bin (P = 0.017)	Bin (P = .002)		Bin (P = 0.047)	Bin (P = 0.001)	V (P = 0.002)	gh (P = 0.08)
							H (P < 0.001)	vh (P = 0.26)
								gvh (P = 0.33)
Trailfoot distance, mm	185 ± 48	199 ± 54	196 ± 51	193 ± 61	204 ± 59	194 ± 55	G (P = 0.83)	gv (P = 0.43)
							V (P = 0.13)	gh (P = 0.78)
							H (P = 0.16)	vh (P = 0.16)
								gvh (P = 0.10)
Penultimate step length, mm	669 ± 134	656 ± 106	633 ± 132	596 ± 139	578 ± 138	552 ± 137	G (P = 0.08)	gv (P = 0.87)
		ND (P = 0.005)	Bin (P < 0.001)		ND (P = 0.005)	Bin (P < 0.001)	V (P < 0.001)	gh (P = 0.022)
			D (P = 0.005)			D (P = 0.005)	H (P < 0.001)	vh (P = 0.40)
								gvh (P = 0.60)
Crossing step length, mm	768 ± 66	737 ± 91	755 ± 58	727 ± 83	721 ± 76	733 ± 79	G (P = 0.29)	gh (P = 0.30)
							V (P = 0.08)	gv (P = 1.00)
							H (P = 0.09)	vh (P = 0.55)
								gvh (P = 0.75)
Walking velocity, mm/sec	1269 ± 126	1253 ± 125	1236 ± 118	1229 ± 194	1192 ± 198	1161 ± 198	G (P = 0.35)	gv (P = 0.17)
		ND (P = 0.005)	Bin (P < 0.001)		ND (P = 0.005)	Bin (P < 0.001)	V (P < 0.001)	gh (P = 0.003)
			D (P = 0.005)			D (P = 0.005)	H (P < 0.001)	vh (P = 0.30)
								gvh (P = 0.17)

 Data are expressed as the mean ± SD. ANOVA main effects are indicated for viewing condition (V): dominant (D) eye viewing; nondominant (ND) viewing; Bin, binocular viewing; H, obstacle height; G, group (visual normal or DS). Interactions between terms are indicated by lowercase letters. Boldface type indicates statistically significant effects.

*  Within cells, Bin, D, and ND indicate results of post hoc analysis, displaying the viewing condition that yielded significantly different results. For example, in the case of VTC in visually normal subjects and DS individuals, VTC was significantly different in D and ND eye (P = 0.001) viewing, relative to habitual viewing (P < 0.001), and significantly different in dominant eye compared with nondominant eye viewing (P = 0.001).

Figure 2.

 

The effect of obstacle height on VTC (both in mm) for the visually normal (VN) and deficient stereopsis (DS) groups for each of the three viewing conditions (Bin, both eyes open; D, dominant eye viewing; ND, nondominant eye viewing). Error bars, ±1 SD of the mean.

Figure 3.

 

The effect of obstacle height on penultimate step length (both in mm) for the visually normal (VN) and deficient stereopsis (DS) groups for each of the three viewing conditions (Bin, both eyes open; D, dominant eye viewing; ND, nondominant eye viewing). Error bars, ±1 SD of the mean.

Adaptive Gait Performance with One versus with Two Eyes

Viewing condition had a significant main effect across groups on VTC (P < 0.001), lead_foot distance (P = 0.002), penultimate step length (P < 0.001) and walking velocity (P < 0.001). Our findings indicate a progressive effect, with performance affected (relative to binocular viewing) under dominant-eye viewing conditions but more affected when viewing with the nondominant eye (Table 2). 

Given that both normal subjects and DS individuals showed a significant increase in VTC and a significant decrease in lead_foot distance when viewing with the dominant eye, relative to binocular viewing, our results suggest that the nondominant eye in the DS group makes a useful contribution to the completion of the task of obstacle-crossing. 

Discussion

Our results show that stereo-deficient individuals were more cautious during adaptive gait involving obstacle-crossing when compared with visually normal subjects, but this only became apparent when the subjects were negotiating higher obstacles. As in visually normal subjects, closing either eye had an effect on performance, suggesting that both eyes in DS individuals contribute to execution of the task. Before discussing these results, we consider the visual demands associated with the task of obstacle-crossing. 

Visual Information Relevant to Obstacle-Crossing

When an obstacle in the travel path has to be negotiated, research suggests that the task consists of two phases, the approach to the obstacle and the stepover (crossing). ^2,6 Two pieces of information are critical for successful obstacle negotiation—namely, updated distance to the obstacle and obstacle height. ²⁴ Previous work investigating gaze behavior in normal subjects show that very few gazes are made at the obstacle during the approach and that none are made during the crossing. ^2,25 This implies that visual information derived from brief viewing is used in a feed-forward manner to estimate obstacle distance and height ² and subsequently to ensure that the feet are appropriately placed before the obstacle and to plan and control limb trajectory over it. As well as feed-forward information, visual information can be used in an online manner to fine-tune movements (e.g., to update foot placement and/or update clearance over the obstacle ²⁶ ). Since gaze is directed around two steps ahead during adaptive gait, ² any online information during the crossing is provided via peripheral visual cues. Another way in which the visual information needed for adaptive locomotion can be considered relates to whether it is exteroceptive (i.e., relating to the environment; e.g., information about the size of the object to be avoided) or exproprioceptive (i.e., relating to the position of the body in the environment). Recent research suggests that visual exteroceptive information is mainly used in a feed-forward manner whereas, visual exproprioceptive information is used online to fine tune movements. ²⁶  

Although the importance of acquiring binocular vision in prehension tasks has been studied extensively, ^27–32 comparatively little attention has been paid to its role in adaptive gait. Patla et al. ⁶ were the first to compare obstacle-crossing performance in monocular and binocular viewing conditions. They found that toe clearance was significantly higher when one eye was occluded but no other gait variable was affected. However, this higher toe clearance occurred only when one eye was occluded during the approach. There was no detrimental effect of switching to monocular viewing during the stepover. The implication of these findings is that monocular vision does not provide adequate visual exteroceptive information, and, as argued by Patla et al., ⁶ the uncertainty about the height of the object leads to increased toe clearance over it because of the need to maintain margins of safety. In more recent work, Vale et al. ⁸ have shown that toe clearance when walking onto a raised surface increases with increasing levels of monocular blur (with resulting reductions in stereoacuity), again suggesting that normal binocular vision during the approach is necessary to extract adequate visual exteroceptive information. 

Effects of Increases in Obstacle Height

The novel aspect of our results relates to the differences found between DS individuals and visually normal subjects which became apparent as the height of the obstacle increased. In their examination of gaze behavior in visually normal subjects, Patla and Vickers ² found that the number of gazes at the obstacle increases as obstacle height increases from 1 to 30 cm. Presumably, this finding reflects the greater importance of acquiring accurate visual information with increasing task difficulty. Our finding that gait parameters were changed to a greater extent in DS individuals, as obstacle height was increased, suggests either that they did not increase their frequency of gaze at the obstacle, or that, if they did, they did not acquire sufficient visual information to allow them to cross the obstacle without increasing safety margins. We did not assess the gaze behavior of our participants in this study, but it represents an obvious area for further research. 

Studies in which the effects of changing obstacle height were investigated in visually normal subjects have reported very different findings, with some studies finding increased toe clearance with increasing obstacle height, ² and others finding the reverse. ²⁴ Others have ignored the effects of changing obstacle height, because the effects were found to be constant across the different vision conditions being investigated. ²⁶ In visually normal subjects, toe clearance when stepping onto a raised surface has been found to decrease with increasing surface height, when vision is bilaterally ^9,33 or unilaterally ⁷ blurred. This reduced clearance as step height increases has been interpreted as an energy conservation strategy, ⁹ since more energy is expended in lifting a limb higher. In the present study, the drive to minimize energy evident in the visually normal subjects when negotiating higher obstacles was not present in the results for our DS participants. This finding strongly suggests they may find obstacle height judgments problematic and that, to reduce the risk of tripping, they increase margins of safety. To give an idea of what aspects of gait were controlled by the DS group to maintain or increase toe clearance margins, we retrospectively determined the correlations between VTC and other relevant gait parameters (across all conditions and for all DS individuals). This analysis highlighted that both leading-foot positioning and penultimate step length covaried with VTC (R ² = 0.11 and 0.12 respectively, P < 0.01). In contrast, trailing-foot positioning was the only measure that covaried with VTC in visually normal subjects (R ² = 0.16, P < 0.001). This additional analysis suggests that visually normal subjects were able to plan in advance the final trailing-limb foot placement needed to achieve adequate toe clearance during leading-limb crossing, whereas DS individuals appeared to be uncertain about the optimum final foot placement and as a consequence altered their penultimate step to facilitate increasing toe clearance during the subsequent crossing step. Possible causes of the DS group's altered gait performance are discussed in a later section. 

Effects of Closing One Eye

When viewing changed from binocular to monocular, adaptive gait in both groups was significantly affected, with vision main effects for several parameters (Table 2). These results differ from those reported by Patla et al. ⁶ who showed that in visually normal subjects, only VTC was affected by monocular occlusion. However, they are broadly consistent with those of Vale et al. ^7,8 and Hayhoe et al., ³⁴ who, as well as increased VTC, also reported a slower walking velocity under monocular conditions in visually normal subjects. 

Dominant versus Nondominant Eye Viewing.

Gait differences due to occlusion of one eye were more numerous and more pronounced under nondominant eye viewing than under dominant-eye viewing conditions (Table 2). These findings are in agreement with previous work in visually normal subjects indicating that monocular blur of the nondominant eye causes VTC when walking onto a raised surface to be increased by a lesser amount than when the dominant eye is blurred. ⁷ In the present study, the pattern of results for dominant and nondominant eye viewing did not differ in the DS group by comparison with that in the visually normal group. A likely explanation for the fact that adaptive gait in nondominant eye viewing was similar to that in visually normal subjects, is that only 9 of the 16 DS individuals had amblyopia; thus, 7 of the DS individuals (i.e., almost half) had a clear view when viewing with the nondominant eye only. Another possible explanation is that CS in nondominant eye viewing was not degraded to the same extent as VA (Table 1). Previous studies have highlighted the greater importance of CS relative to VA for controlling gait and posture. ^35–38  

Contribution of the Ambylopic/Nondominant Eye to Adaptive Gait Performance.

Our results indicate that gait during obstacle-crossing was significantly altered when one eye was occluded in DS individuals, even if the eye occluded was the amblyopic/nondominant eye (Table 2). Therefore, by implication, the nondominant eye in these individuals contributes usefully to the execution of adaptive gait in binocular viewing. This strongly suggests that DS individuals exhibit a degree of binocular cooperation that is not captured by the routinely used static, clinical measures of stereopsis. There are several reports that the processing of the dynamic element of binocular depth perception can be preserved in individuals who lack static stereopsis, ^39–42 and thus we cannot rule out the possibility that such binocular cooperation would have been revealed had measures of dynamic disparity processing (motion stereopsis) been recorded. 

Possible Causes of the Altered Performance in DS Individuals in Binocular Viewing

Despite evidence indicating that the amblyopic eye contributes usefully to adaptive gait performance, the fact remains that, across all viewing conditions, the adaptive gait in DS individuals became more cautious with increasing obstacle height, which was the case, even under habitual viewing conditions, although to a slightly lesser extent. This section considers the possible reasons for this altered performance. 

Reduced or absent stereoacuity represents one obvious reason that our DS individuals behaved differently relative to visually normal subjects as the obstacle height was increased. The importance of stereopsis for movement control remains the subject of considerable debate. ³⁴ For example, McKee et al., ⁴³ suggest that stereopsis may be too crude to be useful in locomotion tasks and that it is primarily used for online control of the hand which, of course, remains in view during movement. In adaptive gait, however, the obstacle or feet are not viewed centrally during the crossing, ^2,25 and thus vision appears to be mainly used in a feed-forward manner. The lack of group differences in leading- and trailing-foot placement before the obstacle but the increased clearance over it as obstacle height was increased that we found in our DS individuals suggests that stereopsis is important for accurate judgments about obstacle height but not about its location. This is consistent with the results obtained by Patla et al. ⁶ However, unlike Patla et al. ⁶ who reported differences only in toe clearance, when stereopsis was degraded by monocular occlusion, we found differences in other gait parameters (i.e., walking velocity and penultimate step-length) for our DS group, but these additional changes may be secondary to the increases made to toe clearance. 

While stereoacuity is known to provide extremely useful depth information it is by no means the only way to extract such information. As well as stereoacuity, vergence provides a binocular cue to distance, and there is a long list of distance cues that are available under monocular conditions that can be used to resolve depth differences in the environment and to identify surfaces (e.g., top surface of obstacle). In relation to the task of obstacle-crossing, the most useful of these may be motion parallax. Gibson ⁴⁴ developed an ecological approach to perception in which movement of the observer plays a vital role. Specifically, flow patterns in the optic array signal the direction of movement and changes in movement direction. Gibson's view was that an observer perceives the surrounding world by actively sampling the optic array to detect invariant information. ⁴⁴ Motion is fundamental because without it, variant information cannot be distinguished from invariant information. Patla and Vickers ² emphasized the importance of optic flow for acquiring self-motion information, which they suggested was, in turn, used to regulate the velocity of locomotion. The cues proposed by Gibson that are used to perceive the world are generally held to be available monocularly. ⁴⁵ However, Jones and Lee ⁴⁶ proposed that the visual system is able to make use of matching information from the two eyes (binocular concordance) in the monocular optic arrays. In fact, they suggested that the ecological benefit of binocular vision may have more to do with binocular concordance than with binocular disparity. If they are correct, poor or absent binocular concordance could be an alternate explanation for the differences we observed between our DS and visually normal individuals other than stereo deficiency. 

Thus, although the importance of binocular vision is not in doubt in adaptive gait, the question of what it is about the binocular vision of DS individuals that makes their behavior differ from that of visually normal subjects is as yet unresolved. Although it may have to do with poor or absent stereoacuity, it could be due to an abnormal combination of information from monocular cues. It is also possible that both explanations are correct. The latter is supported by recent anecdotal reports in which it is claimed that the perceptual salience of so-called monocular cues (e.g., motion parallax) also increases when stereopsis is recovered. ⁴⁷  

Relevance of the Current Findings in the Wider Context: Impact of Amblyopia and Strabismus on Everyday Functioning

In the past decade, considerable attention has been paid to examination of the impact of amblyopia and strabismus on the lives of the individuals who exhibit these conditions. Interest in this topic followed the publication of an influential U.K. report ⁴⁸ of an examination of the literature relating to humans with amblyopia, refractive error, or noncosmetically obvious strabismus. The authors concluded that “there is insufficient evidence in the literature to draw any firm conclusions about the impact of these conditions on quality of life.” Similar views continue to be expressed. ⁴⁹ The report by Snowden and Stewart-Brown ⁴⁸ led to research aimed at understanding the impact of amblyopia and strabismus. This research took three broad approaches: Questionnaire studies ^50–52 and population studies ^14,53 represented two approaches to the examination of how amblyopia and strabismus may affect the lives of individuals with those conditions. A third approach was centered on the functional impact that these conditions may exert. Investigators in studies adopting this approach have examined whether individuals with amblyopia and/or strabismus perform any worse on everyday, real-world tasks than do those classed as being visually normal. Tasks that have been studied include reading, ^54,55 driving, ^56–58 and the visuomotor control of prehension. ¹⁹ In relation to driving, for example, individuals with defective stereopsis were found to perform less well relative to visually normal subjects on a task that involved driving through a slalom course ⁵⁶ and, more recently, Tijtgat et al. ⁵⁷ have shown that a lack of stereopsis is associated with more prudent braking behavior. Whether the differences in driving performance between individuals with and without stereopsis leads to an increased accident rate remains the subject of debate. They suggested that the observed different braking pattern in stereo-defective individuals would not lead to an increased risk of rear-end collisions. Maag et al., ⁵⁸ on the other hand, found that the number of crashes was significantly larger among taxi drivers with binocular vision problems compared with visually normal taxi drivers, but that such crashes were no more severe than those occurring in visually normal subjects. In relation to reaching and grasping, there is evidence from visually normal subjects and from individuals with long-standing stereo deficiency that binocular vision is particularly important for the terminal reach and grasping phases when online control is required and less important for the planning and initial reach phases. ^17,18 However, given the nature of the difference between adaptive gait and reach/grasping, the relative importance of online control in the two tasks may be considerably different. 

The present study represents, to our knowledge, the first examination of how adaptive gait involving obstacle-crossing may be affected in individuals with long-standing stereo deficiency due to amblyopia and/or the conditions with which it is associated. Given the relatively short time during which studies of this nature have been performed and the likelihood that amblyopia and its associated conditions may be expected to affect performance on some locomotion tasks more than others, it is not possible at this point to be certain about their precise impact on all aspects of gait. However, our results indicating adaptive gait involving obstacle-crossing was more cautious in DS individuals when negotiating higher obstacles would suggest they may be at greater risk of tripping and falling in everyday locomotion. Evidence to support this claim comes from numerous previous studies that have consistently found an association between bilaterally reduced VA, the presence of cataract and/or reduced CS, reduced stereoacuity, or increased asymmetry in right versus left eye VA and an increased risk of falling. ^59–63  

Footnotes

 Supported by an RCUK (Research Councils, UK)Academic Fellowship (JGB).

Footnotes

 Disclosure: J.G. Buckley, None; G.K. Panesar, None; M.J. MacLellan, None; I.E. Pacey, None; B.T. Barrett, None

The authors thank David Elliott for his comments on an early draft of the manuscript.