In gaining insights into the role of peripheral visual cues in gait, previous research has determined mobility performance in patients affected by retinitis pigmentosa or glaucoma [both resulting in peripheral field loss (PFL)]. Findings highlight that when negotiating an obstacle course, patients with PFL have a reduced walking speed and experience more obstacle hits compared with age-matched normal sighted subjects.1–3 Furthermore, PFL patients use eye movements to scan more in the vertical direction than normally sighted subjects, likely as a safety strategy to ensure any obstructions at head level (e.g., tree branches, street signs) or on the floor (e.g., surface height changes) are detected well in advance.4 Few studies have investigated the importance of various parts of the visual field to mobility. Lovie-Kitchin et al.5 concluded that the left and inferior midperipheral areas seem to be the most important, and Turano et al.6 found that the central 20° and the lower peripheral field were most important to walking speed and that the loss of the central 20° led to the most obstacle contacts. Lovie-Kitchin et al.5 suggested that this area of research might be better investigated using young subjects and simulated visual field loss while standardizing other visual functions.
The aim of this study was to determine the relative importance of visual cues available from different parts of the peripheral visual field in controlling adaptive gait when stepping over an obstacle. We used three-dimensional motion analysis techniques, which are regarded as the gold standard measurement of gait,7 and simulated visual field loss using goggles, so that we could control the extent and area of the loss and the results were not influenced by other aspects of visual function (visual acuity, contrast sensitivity, and stereopsis) loss that is found in patients with PFL. Previous studies investigating the effects on gait of simulated visual field loss in otherwise normally sighted participants have only investigated the effects of lower visual field occlusion (LO),8–11 and they found that the lower visual field is used for monitoring online foot placement before obstacles and foot trajectory over them. In this study we also investigate the effects of upper visual field occlusion (UO) and circumferential peripheral visual field occlusion (CPO). The obstacle to be negotiated was either positioned as a lone structure or within a doorframe. We included a doorframe around the obstacle to provide positional cues in all parts of the peripheral field. This further extends the work of Rietdyk and Rhea,9 who used vertical positional cues in their study of how vision is used in adaptive locomotion. The top part of the doorframe provided a common real world object that would be seen in the upper visual field.
By manipulating the amount of visual field occlusion, the visual cues from the doorframe and/or obstacle were different in each experimental condition and available at different stages in the approach in the following manner. With CPO, the doorframe and obstacle disappeared from view from approximately two walking step lengths away. With LO, the upper section of the doorframe was always visible, whereas the obstacle disappeared from two walking step lengths away. Finally, with UO, the lower part of the doorframe and obstacle were always visible, whereas the upper section of the doorframe disappeared from two walking step lengths away. Cues were classified as visual exteroceptive when representing static properties of the environment, including doorframe, obstacle, and initial position of subject, and as visual exproprioceptive when related to the dynamic spatial relationship between subject and doorframe/obstacle.
Twelve participants (7 men, 5 women; mean age, 26.16 ± 6.01 years) volunteered to participate. All subjects had visual acuity of 0.0 logMAR (Snellen equivalent 6/6) or better. Ametropic subjects were only included if they habitually wore contact lenses, becuase spectacles could not be worn with the PFL goggles. All participants selfreported that they were healthy and had no current injuries, history of balance, musculoskeletal problems, epilepsy, or migraine. All subjects were asked to wear shorts, T-shirt, and flat soft-soled shoes, so that their gait would be unencumbered. The tenets of the Declaration of Helsinki were observed, and the experiment gained approval from the local Bioethics Committee.
Plain eye-protective goggles were used to provide the four visual field conditions: UO, LO, CPO and full vision (FV) as the control condition.12 These restrictions were applied to the dominant eye, with vision of the nondominant eye being totally occluded. The use of monocular occlusion ensured that the potential influence of variable impairment of stereopsis (provoked by any misalignment across the two eyes if a binocular PFL goggle had been used) was avoided. This approach has been used for the same reason in studies investigating visual field contribution to the control of prehension13 and central visual field contribution to mobility performance.14 The nondominant eye was completely occluded by applying black tape over the corresponding side of the goggles, and the dominant eye (determined using the Kay pictures dominance test, a sighting test) was partly occluded (restricted) in the following manner: upper and lower visual fields were occluded by placing black tape with the upper or lower edge level with the midpoint of the subject's pupil (Fig. 2). CPO was achieved by placing black cardboard with a single hole in front of the pupil, leaving available only the central 20° of visual field.15 We chose to occlude all but 20° of visual field to ensure that the image of the doorframe and obstacle fell within the central visual field from approximately two walking step lengths away from the doorframe/obstacle, because it has been suggested that, during locomotion, exteroceptive information regarding an obstacle in the pathway is obtained about two steps before reaching it.16 The diameter of the hole required to provide a 20° visual field in CPO was determined after measurement of a subject's vertex distance with the goggles.12 The hole was positioned on the goggles by centering the hole while the participants fixated the obstacle and doorframe from two steps away.
Visual acuity was measured with the Early Treatment Diabetic Retinopathy Study logMAR charts (luminance 160 cd/m2) at a distance of 4 m, using a by-letter scoring system. Contrast sensitivity was measured using the Pelli-Robson chart at 1 m (luminance 200 cd/m2), using a by-letter scoring system. For both measurements, different charts were used in random order across visual conditions and subjects to avoid subjects memorising the letters. Mean visual acuity scores for FV and CPO conditions (the least and most perturbed visual conditions, respectively) were −0.10 ± 0.06 and −0.09 ± 0.07 logMAR, respectively (Snellen equivalent 6/5; two tailed t-tests, p = 0.38), and mean contrast sensitivity scores were 1.78 ± 0.15 and 1.72 ± 0. 11 logCS, respectively (two tailed t-tests, p = 0.12). An Esterman monocular visual field test was undertaken on one subject and confirmed that the goggles used for each visual condition occluded the expected extent of the visual field.
We demarcated a walking corridor in our gait laboratory by positioning parallel gray boarding (1.8 m high) 4 m apart over a 7 m length of the laboratory. This ensured that visual cues were consistent across trials. Subjects were asked to look straight ahead (this meant that the obstacle and doorframe would have been seen by the central visual field during the approach to the obstacle, up to about two steps away, but not during crossing17) and start walking from the middle of one end of the corridor at their customary speed and negotiate an obstacle located ∼3 m from the start. Two different obstacle heights (4 and 8 cm) were used and reflected the range of heights of the lower section of typical doorframes. Each obstacle was positioned either as a lone structure or in a way that it formed the bottom section of a doorframe. The doorframe was 212 cm in height and 95 cm in width and had been painted matt white. The obstacle(s) had also been painted matt white and had a luminance of 60.9 cd/m2, with a surround floor luminance of 17.9 cd/m2. The room was illuminated by six fluorescent light fixtures (size 120 × 57 cm) embedded in the ceiling and equally distributed across it. Subjects could not perceive a shadow of the obstacle/doorframe.
Starting positions were determined by asking subjects to walk up to and over one of the obstacles from at least four to five walking steps away. Each subject's starting position was adjusted until they stepped over the obstacle consistently with the same leg in a natural and comfortable manner. Two other starting positions, 20 cm in front or behind the chosen starting point, were included to prevent subjects simply using a motor strategy to negotiate the obstacle, and these starting points were randomized across trials. Trials were completed in two blocks (doorframe with obstacle and obstacle only) and were repeated six times. The four visual conditions, two blocks, two obstacle heights, and six repetitions were completed in random order for a total of 96 trials.
A motion capture system (Vicon MX, Oxford Metrics, United Kingdom) with eight wall or ceiling mounted cameras was used to record (100 Hz) three-dimensional body segment kinematics. Reflective markers were placed on the head (anterio- and posterior-lateral aspects), trunk (jugular notch, sternum, and vertebrae C7 and T10), pelvis (anteriosuperior iliac spines and sacrum), lateral aspects of thighs, knees (femoral condyles), shanks, and ankles (malleoli), and on the feet (end of second toes, second and fifth metatarsal heads, and calcanei). Two additional markers were placed on the upper front edge of the obstacle. Each subject's height, mass, leg length, and knee and ankle width were measured using standard techniques.
By using PlugInGait software (Oxford Metrics), marker trajectory data were filtered using the Woltring spline smoothing routine, with mean square error “filter options” set to “10.” They were then processed to define a three-dimensional link-segment model of each subject, incorporating the individual anthropometric measures taken. A virtual marker, representing the inferior tip of the shoe (virtual shoe tip), was determined by reconstructing its position relative to the markers placed on the second and fifth metatarsal heads and end of second toe. The three-dimensional coordinate data of the sternum, head (×4), each of the foot markers (including the virtual shoe tip), and the markers placed on the obstacle were exported in American Standard Code for Information Interchange format for further analysis.
Analysis focused on foot placement parameters during the approach and toe clearance parameters over the obstacle. Head movements were also analyzed to confirm whether the subjects looked straight ahead during obstacle crossing as requested.
The dependent measures analyzed were as follows (Fig. 1):
* Lead-limb toe clearance (mm): the vertical distance between the virtual shoe tip marker and the upper edge of the obstacle at point of crossing.
* Lead- and trail-foot horizontal distance before obstacle (mm): the anterioposterior horizontal distance between the position of the virtual shoe tip (during ground contact) and the front edge of the obstacle.
* Variability across repetitions of lead-limb toe clearance and of lead- and trail-foot horizontal distance.
* Crossing-walking velocity (mm/s): the mean instantaneous velocity of sternum marker in anterioposterior direction, from lead-limb heel contact before obstacle to lead-limb heel contact after the obstacle.
* Range in head flexion (sagittal plane; deg): from lead-limb heel contact before obstacle to lead-limb heel contact after the obstacle.
* Range in vertical head movement (mm): from lead-limb heel contact before obstacle to lead-limb heel contact after the obstacle.
Statistical analysis was performed with SPSS 15.0 (LEAD Technology). Dependent measures were tested for normality with the Kolmogorov-Smirnov test (p level set at 0.05). Data were normally distributed for 89 of 96 distributions for lead-foot horizontal distance, 86 of 96 distributions for trial-foot horizontal distance, 87 of 96 distributions for lead toe clearance, 93 of 96 distributions for crossing-walking velocity, 79 of 96 distributions for head flexion, and 95 of 96 distributions for vertical translation. Because, the majority of the data sets within each dependent measure were normally distributed for each dependent measure, repeated-measures four-way analysis of variances were used to determine the effects of the following:
* Visual field condition: four levels (UO, LO, CPO, and FV);
* Obstacle type: two levels (obstacle with doorframe, obstacle only);
* Obstacle height: two levels (8 and 4 cm);
* Repetition: six levels.
Post hoc analyses were undertaken using Tukey's Honestly Significant Differences test. For the variability of the dependent measures, data were normally distributed in 15 of 16 distributions for lead- and trail-foot horizontal distance, whereas for lead-vertical toe clearance, all the data sets were normally distributed. Repeated-measures three-way analysis of variances was used to determine the effects of visual field condition, obstacle type, and height. The significant p level was set at 0.05 in all cases.
Head flexion and vertical translation were unaffected by all factors (p = 0.38). The average head flexion for each visual condition was compared with the average head flexion maintained during a static standing calibration trial recorded for each subject before data collection to rule out the possibility that participants walked with their head constantly flexed across all trials. Differences from the calibration trial were in the order of a few degrees: 2.20° in CPO, 2.06° in UO, 3.27° in LO, and −1.23° in FV conditions (negative numbers refer to extension of the head compared with the calibration trial), and they were not significant between conditions (p = 0.34).
Repetition significantly increased lead-foot horizontal distance, crossing-walking velocity and significantly decreased lead-limb toe clearance (p = 0.043), but there was no significant interaction between repetition and either visual field condition or obstacle type (p = 0.1).
Foot Position Before the Obstacle
Lead-foot horizontal distance was significantly affected by visual field condition (p = 0.001, Fig. 2a) and was greater in CPO and LO conditions than in UO or FV (post hoc, p = 0.033). There was a significant interaction between visual field condition and obstacle type (p = 0.033), and post hoc analysis highlighted that LO, UO, and FV showed no significant differences in lead-foot horizontal distance with the doorframe present, whereas it remained larger in CPO (p = 0.012).
Trail-foot horizontal distance was significantly affected by visual field condition (p = 0.001, Fig. 2b) and was greater in CPO compared with UO, FV (post hoc, p = 0.001), or LO conditions (post hoc, p = 0.042). A significant interaction between visual field condition and obstacle type was observed (p = 0.008), and post hoc analyses showed that the increase in trail-foot horizontal distance in the CPO condition increased further with the doorframe (p = 0.027) compared with the obstacle only. Unlike with lead-foot horizontal distance, trail-foot distance under LO conditions with the doorframe was significantly greater than under FV conditions with the obstacle only (p = 0.035).
Lead-limb toe clearance was significantly affected by visual field condition (p = 0.001) and was greater in CPO and LO compared with UO or FV conditions (Fig. 2c). A significant interaction between visual field condition and obstacle type was observed (p = 0.002), and post hoc analyses showed that the increase in lead-limb toe clearance in CPO was greater with the doorframe (post hoc, p = 0.001). Lead-limb toe clearance was greater for the low compared with high obstacle (p = 0.001).
Crossing-walking velocity was significantly affected by visual field condition (p = 0.001), and was reduced in CPO and LO compared with UO or FV conditions (Fig. 2d). A significant interaction between visual field condition and obstacle type was observed (p = 0.005), and post hoc analyses showed that crossing-walking velocity was further reduced in CPO when the doorframe was present (p = 0.022).
Variability in lead-foot horizontal distance was not influenced by any factor (p = 0.17). A significant effect of vision condition was found for variability in trail-foot horizontal distance (p = 0.01) with higher variability for CPO and LO. Variability in lead-limb toe clearance was higher in CPO and LO conditions compared with UO or FV conditions (p = 0.026; Fig. 3).
Head flexion range and vertical translation were constant across the visual conditions, and average amount of head flexion throughout the trials was similar to the calibration static trial head position. These results indicate that subjects followed the instruction to look straight ahead and did not alter the visual field available by moving their head.
Lead-limb toe clearance was significantly reduced when negotiating the higher obstacle. This finding is consistent with that found for stepping up and on to raised surfaces of increasing height and has been suggested to be an energy-saving strategy (as increasing limb elevation requires more energy expenditure).8,18 Lead-toe clearance was also significantly reduced for the last three repetitions compared with the first three, suggesting that subjects used feedback, indicating that no tripping had occurred in the early trials to reduce foot clearance. Crossing-walking velocity increased for the last three repetitions, and consequently, lead-foot placement also increased with repetition. This suggests that subjects became more comfortable with the environment as the experiment progressed. The lack of interaction between repetition and visual field condition (p = 0.1) indicates that these effects did not influence the study's main results concerning PFL.
All subjects successfully cleared the obstacle in CPO and LO conditions, suggesting that visual exteroception information of the obstacle provided in a feed-forward manner from central visual field information, was sufficient to plan successful obstacle negotiation. Lead- and trail-foot horizontal distance and lead-limb toe clearance were greater, and crossing-walking velocity was reduced in CPO and LO compared with FV and UO conditions. The lack of any change with UO indicates that these safety-driven adaptations of increasing margins of safety between feet and obstacle to avoid tripping are not because of a loss in visual field per se but indicate that they are because of the lack of online visual exproprioceptive information of the lower limbs.9,10,16
Vertical toe clearance increased by ∼3 cm under LO and CPO compared with the FV condition (Fig. 2c): increasing the distance between foot and obstacle is a well-known safety strategy adopted to avoid trips and falls.9,10,16 The toe clearance increase from about 9 to 12 cm may seem small, but increases seem to be a compromise between safety and energy cost.19,20 Small toe clearances will increase the chance of a trip if toe clearance is only marginally greater than toe clearance variability, but very large toe clearances would increase the energy cost of gait considerably because of having to lift the mass of the entire limb higher. Our data agree with previous literature in that increases in toe clearance parameters during obstacle crossing has typically been found to be in the order of a few centimetres in healthy populations and are usually found to be ∼10 cm.21,22 Foot placement closer to the obstacle can also threaten stability or safety. For instance, if the trail limb is placed at a reduced horizontal distance from the obstacle, this will cause the lead-limb to subsequently cross the obstacle during the early part of its swing phase when there is reduced hip, knee, and ankle flexion; increasing the chance of the toe contacting the obstacle.23
The increased variability in toe clearance and trail foot placement without vision of the lower limbs (LO and CPO) indicates that visual exproprioception cues from the lower visual field are required for “fine-tuning” limb movements after feed-forward exteroception information is used to plan gait. These findings may help to explain the adaptive gait results when stepping up onto a step in well-adapted multifocal lens wearers.8,24,25 When wearing the multifocals, the patients increased their toe clearance and their foot placement before the step compared with that when wearing single vision lenses. In agreement with previous work,9,10 lead-foot horizontal distance with LO returned to “normal” values (i.e., FV condition values) when the doorframe was present, showing that visual exproprioception of head position relative to the doorframe compensated for the lack of visual exproprioception of the lower limb. Lead-limb toe clearance with LO did not return to normal values with the doorframe present likely because the visible part of the doorframe gave information about the horizontal position of the obstacle but not its height, underlining that lead-limb toe clearance is dependent on visual exproprioception of the lower limbs.9
Our results differ from those of Rietdyk and Rhea9 in that they found that trail-foot placement returned to normal values in LO when positional cues were available at the sides of the obstacle. In our study, although the trail-foot position decreased with the doorframe positional cue, it did not reach normal FV values. This may be because of several reasons: (1) our use of monocular conditions rather than binocular, (2) the height of obstacles used (they used 10, 20, and 30 cm9), making the gait tasks in the two studies somewhat different,26,27 and (3) the manner in which visual field was occluded: in our study lower visual field was occluded from the midpoint of each subject's pupil downwards, standardizing the amount of visual cues unavailable among the participants, whereas in previous studies the same occlusion goggles (basket ball training goggles) were used by all the subjects.9,10
UO did not affect any of the parameters measured, and all results under UO were similar to the FV (control) condition. This lack of change with UO suggests that all CPO results should have mirrored those from the LO condition (although note that the loss of the lower visual field in LO was greater than in CPO). Gait adaptations were similar for CPO and LO in the obstacle only condition. This confirms previous studies that highlight the role of exteroceptive cues being used in a feed-forward manner from central visual field information gained during the approach.17 However, adaptive gait changes were different for CPO and LO when the doorframe was added (Fig. 2). Trail-foot placement and lead-limb toe clearance increased further with CPO, and crossing- walking velocity was further reduced, yet, no significant changes occurred with LO. This may have been because of concerns about hitting the doorframe with the head and/or upper body in the CPO condition. In CPO, without the online visual cues provided by lamellar flow12,28,29 and visual exproprioception that were available in the LO condition, feed-forward exteroceptive information was not sufficient during this complex task, and additional safety driven gait adaptations were thus used. This suggests that there is a hierarchy of importance in PFL as regards to gait, with upper field loss alone causing least problems, lower field loss more problems, and circumferential field loss the most problems. The greater importance to mobility of the lower visual field compared with the upper visual field agrees with previous studies in patients with PFL,5,6 but these studies did not examine correlations between mobility and combined areas of the visual field. A hierarchical three-category PFL model may also help explain results in studies that compare the effects of PFL from different regions. For example, Freeman et al.15 found no increased falls risk in patients with just lower or just upper visual field loss but found a significant link when the field loss in the two areas were combined.
In conclusion, we suggest that visual information, available in the central visual field and provided in a feed-forward manner, allowed subjects to safely negotiate the obstacle when online peripheral visual information was unavailable. However, a loss of information from the lower visual field (under CPO and LO conditions) led to reduced online control of the lower limbs, which in LO conditions was compensated to some extent by the presence of the doorframe available in the upper visual field. The similarity of the findings between LO and CPO when stepping over the obstacle suggest that exproprioceptive information is typically provided by the lower peripheral visual field and used for online control, whereas exteroceptive cues are provided by the central visual field and are used for feed-forward planning. The loss of the whole circumferential peripheral field had the greatest impact on gait likely because of the absence of visual exproprioception information provided by the lamellar flow.
J.B. was supported by an RCUK Academic Fellowship (Research Councils, United Kingdom).
Bradford School of Optometry and Vision Sciences
University of Bradford
Bradford, West Yorkshire BD7 1DP
1. Black A, Lovie-Kitchin JE, Woods RL, Arnold N, Byrnes J, Murish J. Mobility performance with retinitis pigmentosa. Clin Exp Opt 1997;80:1–12.
2. Turano KA, Rubin GS, Quigley HA. Mobility performance in glaucoma. Invest Ophthalmol Vis Sci 1999;40:2803–9.
3. Geruschat DR, Turano KA. Estimating the amount of mental effort required for independent mobility: persons with glaucoma. Invest Ophthalmol Vis Sci 2007;48:3988–94.
4. Vargas-Martin F, Peli E. Eye movements of patients with tunnel vision while walking. Invest Ophthalmol Vis Sci 2006;47:5295–302.
5. Lovie-Kitchin J, Mainstone J, Robinson J, Brown B. What areas of the visual field are important for mobility in low vision patients? Clin Vis Sci 1990;5:249–63.
6. Turano KA, Broman AT, Bandeen-Roche K, Munoz B, Rubin GS, West S; SEE Project Team. Association of visual field loss and mobility performance in older adults: Salisbury Eye Evaluation Study. Optom Vis Sci 2004;81:298–307.
7. Black A, Wood J. Vision and falls. Clin Exp Optom 2005;88:212–22.
8. Johnson L, Buckley JG, Scally AJ, Elliott DB. Multifocal spectacles increase variability in toe clearance and risk of tripping in the elderly. Invest Ophthalmol Vis Sci 2007;48:1466–71.
9. Rietdyk S, Rhea CK. Control of adaptive locomotion: effect of visual obstruction and visual cues in the environment. Exp Brain Res 2006;169:272–8.
10. Rhea CK, Rietdyk S. Visual exteroceptive information provided during obstacle crossing did not modify the lower limb trajectory. Neurosci Lett 2007;418:60–5.
11. Marigold DS, Patla AE. Visual information from the lower visual field is important for walking across multi-surface terrain. Exp Brain Res 2008;188:23–31.
12. Graci V, Elliott DB, Buckley JG. Peripheral visual cues affect minimum-foot-clearance during overground locomotion. Gait Posture 2009;30:370–4.
13. Gonzalez-Alvarez C, Subramanian A, Pardhan S. Reaching and grasping with restricted peripheral vision. Ophthalmic Physiol Opt 2007;27:265–74.
14. Pelli DG. The visual requirements of mobility. In: Woo GC, ed. Low Vision: Principles and Applications. New York: Springer-Verlag; 1986:134–46.
15. Freeman EE, Munoz B, Rubin G, West SK. Visual field loss increases the risk of falls in older adults: the Salisbury eye evaluation. Invest Ophthalmol Vis Sci 2007;48:4445–50.
16. Patla AE. How is human gait controlled by vision? Ecol Psychol 1998;10:287–302.
17. Patla AE, Vickers JN. Where and when do we look as we approach and step over an obstacle in the travel path? Neuroreport 1997;8:3661–5.
18. Heasley K, Buckley JG, Scally A, Twigg P, Elliott DB. Stepping up to a new level: effects of blurring vision in the elderly. Invest Ophthalmol Vis Sci 2004;45:2122–8.
19. Armand M, Huissoon JP, Patla AE. Stepping over obstacles during locomotion: insights from multiobjective optimization on set of input parameters. IEEE Trans Rehabil Eng 1998;6:43–52.
20. Rhea C, Rietdyk S. Gait adaptation: lead toe clearance continually decreased over multiple exposures with and without on-line visual information. Poster presented at the XXth Congress of the International Society of Biomechanics, Cleveland, OH, August 2005. Available at: http://www.asbweb.org/conferences/2005/pdf/0550.pdf
. Accessed September 16, 2009.
21. Chou LS, Draganich LF. Stepping over an obstacle increases the motions and moments of the joints of the trailing limb in young adults. J Biomech 1997;30:331–7.
22. Patla AE, Rietdyk S. Visual control of limb trajectory over obstacles during locomotion: effect of obstacle height and width. Gait Posture 1993;1:45–60.
23. Chou LS, Draganich LF. Placing the trailing foot closer to an obstacle reduces flexion of the hip, knee, and ankle to increase the risk of tripping. J Biomech 1998;31:685–91.
24. Menant JC, St George RJ, Sandery B, Fitzpatrick RC, Lord SR. Older people contact more obstacles when wearing multifocal glasses and performing a secondary visual task. J Am Geriatr Soc 2009;57:1833–8.
25. Johnson L, Buckley JG, Harley C, Elliott DB. Use of single vision spectacles improves stepping precision and safety when elderly habitual multi-focal wearers negotiate a raised surface. J Am Geriatr Soc 2008;56:178–80.
26. Patla AE. Understanding the roles of vision in the control of human locomotion. Gait Posture 1997;5:54–69.
27. Warren WH. Action modes and laws of control for the visual guidance of action. In: Meijer OG, Roth K, eds. Movement Behavior: “The” Motor-Action Controversy. Amsterdam: North-Holland; 1988:339–80.
28. Koenderink JJ. Optic flow. Vision Res 1986;26:161–79.
29. Atchley P, Andersen GJ. The effect of age, retinal eccentricity, and speed on the detection of optic flow components. Psychol Aging 1998;13:297–308.