Obstacle crossing is a discrete event, which only occurs sporadically during everyday walking. Successful and safe obstacle negotiation involves appropriate placement of the feet before and after the obstacle, and adequate clearance over it (Patla 1998). Vision is known to be important for this adaptive gait task: for review see (Marigold 2008). Feedforward information gained via central vision is used to plan the appropriate gait adaptations (to enable appropriate foot placements) and determine the obstacle’s height and other important characteristics (exterioceptive information) (Rietdyk and Rhea 2006; Rhea and Rietdyk 2007; Timmis and Buckley 2012; Heijnen et al. 2014). Then, feedback from the lvf is used online to update (‘fine tune’) foot placement relative to, and foot trajectory over, the obstacle (ex-proprioceptive information) (Rietdyk and Rhea 2006; Rhea and Rietdyk 2007; Timmis and Buckley 2012). Hence, when the lvf is occluded and visual ex-proprioceptive information is unavailable, foot placement distance from an obstacle, and the clearance over it, is increased. Such findings have led to the conclusion that lower limb-floor visual ex-proprioceptive information is habitually used online in controlling the foot during obstacle crossing (Patla 1998; Rhea and Rietdyk 2007). Based on the findings from Study One above, it is also possible that the increase in toe clearances that occur following lvf occlusion is, at least in part, an acute response to the visual perturbation. If this is the case, such a response would diminish with time/repetition following a period of adaptation after the introduction of lvf occlusion.

The aim of the present study was to determine if and how increases in foot placement distance before an obstacle, and the clearance over it, that arise as a result of occluding the lvf, diminish over time. We asked participants to walk around the same ‘figure of eight’ walking pathway as used in Study One, but this time an obstacle was placed in their travel path for the duration of the first, fifth and tenth minutes of walking; participants crossed over the obstacle twice during each of these 1 min periods. Participants again completed two 10 min walking bouts, one under full-field conditions and the other with lvf occlusion. If under lvf occluded conditions, the increases in toe clearance over the obstacle returned back to unperturbed levels within approximately 5 min of walking, this would indicate that lower limb-obstacle visual ex-proprioceptive feedback is not paramount to the control of the swing limb/foot, and thus that the induced initial increase in clearance was an acute response to having vision perturbed. However, if toe clearance remained higher than the unperturbed level for the duration of the 10 min walking trial, even if it initially partly diminished towards unperturbed levels, then this would suggest that lower limb-obstacle visual ex-proprioceptive feedback is paramount for the control of the swing limb/foot during adaptive gait involving obstacle crossing.

Eighteen healthy young adults (mean (SD) age; 21.5 (2.5) years, height; 1.77 (0.09) m, mass; 72.6 (13.2) kg, 8 females) participated. None had taken part in Study One. Note, for each study (Study One and Study Two), recruitment numbers were based on convenience samples; no sample size calculations were undertaken a priori. The disparity in sample size between studies is therefore chance occurrence. As in Study One, prior to data collection all participants gave written, informed consent. Ethical approval was obtained from the Health Sciences Research Ethics Board at Queen’s University (approval number: 6013096). The procedures used in this study adhere to the tenets of the Declaration of Helsinki.

The protocol and data collection approach were identical to Study One, except that participants negotiated a 10 cm high, 3 cm deep and 2 m wide wooden obstacle that was intermittently placed in their travel path during the ten minutes of continuous walking around the figure-of-eight walking circuit (Fig. 1). Participants were instructed to walk around the figure-of-eight circuit at their customary walking speed and, whenever an obstacle was present, to step over it. No other instruction was given. The obstacle was placed approximately 1 m past halfway along one of the 9.5 m straights in the figure-of-eight circuit for the duration of the first, fifth and tenth minutes of continuous walking. Each participant crossed the obstacle twice each time during the one minute period when the obstacle was present. Thus, they crossed the obstacle six times in each visual field condition. For each visual field condition, kinematic data were recorded for the periods of obstacle crossing (i.e. from approximately two-to-three steps before the obstacle, and for one step after crossing the obstacle).

For each participant, the following parameters were determined: lead-limb vertical toe clearance (VTC), lead and trail foot placement before the obstacle, lead foot placement after the obstacle, and approach speed to the obstacle. VTC was defined as the vertical distance between the top of the obstacle and the ‘shoe-tip’ when it was directly above the obstacle. Foot placements were defined as the horizontal distance between each ‘shoe tip’ (during ground contact) and the obstacle. Approach speed was defined as the average forwards velocity of the head segment’s centre of mass for the period of ‘straight-line walking’ during the approach to the obstacle up to the instant of VTC (i.e. over approximately 4–5 steps).

Data were normally distributed and hence were compared using repeated measures ANOVAs with visual field (full, lvf-occluded), time period (first, fifth and tenth minute) and crossing occurrence within each time period (first and second crossing) as factors. Post hoc analyses were made using Tukey HSD tests. The alpha level was set at 0.05. All statistical analyses were undertaken using Statistica for Windows version 8.0 (StatSoft, Inc., Tulsa, OK, USA).

Under each visual condition the obstacle was repeatedly crossed without contacting it, with either the leading or the trailing foot. Lead-limb VTC was significantly affected by visual field (p = 0.002), time period (p = 0.011), but not by crossing occurrence (p = 0.86). There were no interactions between terms (p ≥ 0.14). VTC was higher, on average, when the lvf was occluded (13.6 cm) compared to full field condition (10.0 cm). Across both visual field conditions, VTC became reduced with time period, with post-hoc analyses indicating a significant reduction in toe clearance between the first and fifth minutes (by an average of ~ 1.1 cm, p = 0.025), but no difference between the fifth and ten minutes (p = 0.30, Fig. 4c). Further inspection of the data indicated that the reduction in clearance between the first and fifth minutes was greater for the lvf occluded (~ 1.79 cm, p = 0.028), compared to full field condition (~ 0.42 cm, p = 0.097).

Group mean (+ SE) obstacle crossing a) approach speed, b) trail-limb foot placement before obstacle, and c) vertical toe clearance for minutes 1, 5 and 10 during the ten minutes of continuous walking under full field and lvf occluded conditions. Values are average of the two crossing occurrences that occurred within each one-minute period. The symbol *, indicates a significant difference relative to the full-field condition (p < 0.005), and the symbol ^ indicates a significant difference relative to minute-1 (p = 0.028)

Approach speed was significantly affected by visual field condition (p = 0.005), but not by time period (p = 0.82) or crossing occurrence (p = 0.14). However, there was a significant time period-by-crossing occurrence interaction (p ≤ 0.0013). The average approach speed was lower when the lvf was occluded (1.43 m/s) compared to full-field vision (1.48 m/s; Fig. 4a), and was greater, across visual conditions, for the first crossing occurrence compared to the second during the first minute of walking. However, there were no speed differences between crossing occurrences during the fifth or tenth minutes. There was also a visual field-by-time period interaction trend on approach speed (p = 0.08), which indicated that the approach speed reduced slightly from the first (1.5 m/s) to the tenth (1.47 m/s) minute under full field conditions, whereas it increased from the first (1.42 m/s) to the tenth (1.46 m/s) under lvf occluded conditions. This trend indicated that although approach speed was reduced for the lvf occluded compared to full field condition across all time periods, the reduction in walking speed was more obvious for the first minute of walking compared to that in minutes 5 and 10 (Fig. 4a). Lead-foot placement (i.e., penultimate step) before (average: 1.074 m full field, 1.057 m lvf occluded) and placement after (average: 0.318 m full field, 0.323 m lvf occluded) the obstacle were unaffected by visual field, time period, or crossing occurrence (p > 0.24). Trail-limb foot placement (i.e., final step) before the obstacle, however, was affected by visual field condition (p = 0.010) and by time period (p = 0.042), but was unaffected by crossing occurrence (p = 0.22), and there were no interactions between terms (p > 0.21).

Trail-limb foot placement distance was greater when the lvf was occluded (30.26 cm) compared to the full-field condition (26.72 cm), and for both visual field conditions it increased with time period, with post-hoc analyses indicating no significant increases between the first and fifth, or fifth and tenth, minutes (p > 0.36); however, there was an increase between the first and tenth minutes (by an average of ~ 3.0 cm, p = 0.033; Fig. 4b). Further inspection of the data indicated that this increase in placement distance with time period was mainly related to increases for the lvf occluded condition, with little change in placement distance across time periods in the full field condition.