PLoS ONE
Public Library of Science
image
Gaze-behaviors of runners in a natural, urban running environment
DOI: 10.1371/journal.pone.0233158, Volume: 15, Issue: 5, Pages: 0-0
Article Type: research-article, Article History
Abstract

Gaze-tracking techniques have advanced our understanding of visual attention and decision making during walking and athletic events, but little is known about how vision influences behavior during running over common, natural obstacles. This study tested hypotheses about whether runners regularly collect visual information and pre-plan obstacle clearance (feedforward control), make improvisational adjustments (online control), or some combination of both. In this study, the gaze profiles of 5 male and 5 female runners, fitted with a telemetric gaze-tracking device, were used to identify the frequency of fixations on an obstacle during a run. Overall, participants fixated on the obstacle 2.4 times during the run, with the last fixation occurring on average between 40% and 80% of the run, suggesting runners potentially shifted from a feedforward planning strategy to an online control strategy during the late portions of the running trial. A negative association was observed between runner velocity and average number of fixations. Consistent with previous studies on visual strategies used during walking, our results indicate that visual attentiveness is part of an important feedforward strategy for runners allowing them to safely approach an obstacle. Thus, visual obstacle attention is a key factor in the navigation of complex, natural landscapes while running.

Keywords

Introduction

Vision is intricately linked with movement and plays an essential role in guiding locomotion by influencing how an individual reacts to and navigates their environment [113]. Previous research suggests that the neuromuscular system adjusts gait parameters based on visual input through two distinct, interrelated strategies: 1) a “feedforward sample controlled method,” and 2) an “online control method,” [1,2,46, 911, 1317].

The feedforward sample controlled method predominates when objects in the environment are stationary, which allows for the identification of obstacles on a path and provides time for planning and adjusting gait in anticipation of a stimulus [1, 2, 5, 10, 13, 17]. This allows individuals to utilize their prior experience with the obstacle, or obstacles like it, to safely navigate an approach [1,2,5,10,16,18]. The online control method relies more on peripheral, lower-order processing of visual information and is not based on pre-planning [1, 5, 6, 9, 11]. Hence, individuals can make quick, small adjustments in response to rapidly changing stimuli in the environment [1, 6, 9, 11,19]. Although often discussed as separate processes, these two methods are usually utilized in parallel, facilitating safe navigation of obstacles in one’s path. Thus, these two methods allow individuals to regulate step placement on a step-by-step basis and allow them to anticipate and plan a safe path while navigating complex environments [1, 2, 46, 911,13,17,20].

Considerable empirical research has explored the importance of vision and obstacle avoidance during walking and other athletic events [2, 9,1933], but there has been less research dedicated to vision in running. Bradshaw and Sparrow (2001) evaluated gaze during running, using head angle rather than a gaze tracker, where they identified similar visual strategies as seen in walking. Their study described three phases of a run associated with visual input and control: (1) an “acceleration from rest” at the beginning, (2) a “global visual control phase” which is a period of adjustment to gain one’s heading, and (3) a “local visual control phase” that is established once one is at a comfortable speed and direction of travel and allows for fine adjustments as one moves toward a goal or obstacle. Thus, runners start by accelerating, where they use their global visual control to get a broad sense of their surroundings and environment allowing planning for movement patterns, and then use local visual control to focus on objects of interest in their running path and make small adjustments in order to safely and efficiently navigate different obstacles [23]. The global visual control phase and local visual control phase are comparable to the feedforward sample control method and the online control method respectively and suggest that runners, like walkers, utilize different strategies throughout the action of navigating and clearing an obstacle.

Looking more specifically at the approach and clearance of obstacles during walking, Patla and Greig (2006) suggested that, when faced with an obstacle, focused visual input is essential to make adjustments during the approach rather than during the act of clearing the obstacle itself. This pattern has been observed both in the laboratory and in complex, outdoor environments [1, 2, 5, 9,10,13,17,23, 26, 31]. Regardless of the setting, these studies suggest that individuals most effectively traverse an obstacle when they intermittently sample the object of interest multiple times throughout their approach and have increased, more directed, visual attention when they are approximately two step lengths from the obstacle; supporting a feedforward model during one’s approach [2,9,28,29,30]. This demonstrates that, although continuous visual input is important, increased, directed visual attention just prior to approaching an obstacle may also be critical for successful clearance [2, 28, 29, 30, 31].

In the study of peripheral vision, in the context of approaching an obstacle, it plays a role in determining where directed visual attention will be placed. Once this occurs, it appears as is supported above, that directed visual cues from the central visual field predominate [34, 35]. Where peripheral vision appears to dominate is when individuals are clearing an obstacle and transition to “improvisational” online visual control [9,19, 24, 25, 27, 36]. Marigold et al. (2007) observed that when obstacles suddenly appeared within two steps of a person, peripheral vision and an occasional saccade (i.e., a very rapid eye movement directed towards an object) were sufficient to navigate it without the need for a longer span of directed attention on the obstacle. The idea of “attention” was explored further by Weerdesteyn and colleagues (2003, 2005) where they observed higher rates of failure among individuals navigating obstacles while distracted (i.e., multitasking) and in older versus younger individuals when navigating unexpected obstacles.

Additionally, using visual attention to clear obstacles during locomotion becomes more challenging as speed increases. It is thought that as speed increases runners have reduced processing time during their approach, resulting in decreased fluency, stability, and accuracy as they approach an obstacle [22, 37]. Bradshaw and Sparrow (2001) tested the effect that velocity had on visual processing and anticipation by measuring the gait of individuals as they walked, ran, and sprinted toward different obstacles. Their study found that the onset of visual control is directly related to the speed individuals’ run toward an obstacle, demonstrating that sprinting participants had a later onset of visual control. These results coupled with a later experiment on long jumping [26] suggest that when individuals move at higher rates of speed, there may be less visual control and a reduced ability to anticipate hazards in their path.

Visual attention can be defined by determining visual fixation on an object. A fixation is generally taken to represent the intentionality in gaze and is a specific period of time in which an individual obtains relevant information about their environment [2, 38, 39, 40]. Although the time that defines a single fixation remains debated in the literature, with numbers ranging from 80–150 ms, a number of studies using similar methods to ours use a value of 99 ms as representative of intentionality in gaze behavior [2, 38, 39, 40]. Thus, in the current study, we also use 99 ms to quantify the timeframe of a fixation, which allowed us to explore the role of vision during the approach of an obstacle (i.e., a sidewalk curb) while running in an outdoor environment.

Specifically, this study attempts to answer the question, “Do runners fixate on a sidewalk curb when running in an urban setting?” To address this question and to focus our investigation, this study utilized a gaze-tracker and a custom-made algorithm that allows for comparisons between computer and human analyses of data to explore two primary hypotheses (our null hypotheses) and their alternates.

Hypothesis 1

H1o

The majority of runners in our study will use an online sampling method as evidenced by the absence of fixations on the obstacle throughout the entirety of the run.

H1a

The majority of runners in our study will use a feedforward sampling approach as evidenced by fixations on the obstacle throughout the approach phase of their run as they move toward the obstacle (i.e., the sidewalk curb).

H1b

The majority of runners in our study will utilize a combination of feedforward and online sampling strategies as indicated by the occurrence of interspersed fixations on the obstacle with times of no fixation on the obstacle.

Hypothesis 2

H2o

The average velocity of the runner has no impact on the number of fixations on the obstacle along the run.

H2a

The faster a runner travels the fewer fixations on the obstacle throughout a run, and thus the slower a runner travels the more fixations on the obstacle throughout the run.

Methods

Experiment, participants, and equipment

Ten runners (5 males and 5 females; Leg Length 95.19 +/- 7.37-cm; Table 1) were recorded running on a 20-m long sidewalk path with an obstacle (i.e., a curb that was 0.15-m high), a natural, urban running environment (Fig 1). Participants wore a gaze-tracker, modified with an adjustable visor (Zhi Jin, Hong Kong, CN), and its small backpack (CamelBak, Petaluma, CA, USA) while jogging or warming up in their normal manner (Fig 2). Running with a gaze tracker has the potential to induce movement that can affect crosshair position and cause crosshair drop-outs. To minimize this effect the shield and the head band secured the glasses portion of the unit securely to the runner’s face, limiting its motion.

Natural running environment showing Curb: This is a natural running environment in a typical city in the United States.
Fig 1
The track was made of concrete. A) The track participants ran as they approached the curb. B) Arrows designate the curb participants cleared. The cone in the background designated the turning point for runners to start another lap.Natural running environment showing Curb: This is a natural running environment in a typical city in the United States.
ISCAN Omniview-TX Head-Mounted Eye tracking device worn by one of our authors (MC).
Fig 2
This has been modified by combining it with a face shield (portion in yellow), allowing the glasses to be held tightly against the face even while moving. In addition, the transmitter was placed into a backpack, which allowed for a natural place for the transmitter to be housed while an individual was running.ISCAN Omniview-TX Head-Mounted Eye tracking device worn by one of our authors (MC).
table-wrap
Table 1
Demographic characteristics and average running velocity for the subject pool used in this study (n = 10) as an aggregate and divided by gender.
Characteristic Sample (n = 10) Range Male (n = 5) Female (n = 5)
Age (yrs.) 22.50 +/- 2.76 21–29 22.80 +/- 3.49 22.20 +/- 2.17
Height (m) 1.74 +/- 0.12 1.55–1.93 1.84 +/- 0.06 1.64 +/- 0.08
Weight (kg) 75.98 +/- 16.97 52.16–115.67 85.28 +/- 17.54 60.33 +/- 11.16
Velocity (m/s) 3.02 +/- 0.14 2.81–3.18 3.06 +/- 0.14 2.97 +/- 0.15
Experience (yrs.) 1 5.60 +/- 2.91 1–7 5.00 +/- 2.35 6.20 +/- 3.56

Data were collected as participants ran at a self-selected distance-running pace and took place at dawn and dusk to limit light exposure. At approximately 15-m into the run, participants encountered and cleared the curb. Participants ran up to 15 laps along the sidewalk and in a loop back to the start of the path, with each lap considered a “trial.” All protocols were approved by the Duke University IRB (protocol 2017–0947) and the individual in this manuscript (Fig 2) has given written informed consent (as outlined in the PLOS consent form) to publish these case details.

The current study uses methods similar to previous studies including the use of eye tracker equipment [2, 3, 11, 33], and the use of a natural environment to study human locomotion [33]. The gaze-tracker (Omniview-TX Head-Mounted Eye Tracking, 30 Hz, ISCAN Inc., Woburn, MA, USA) was composed of three different cameras. The two nearest the eyes measured the right and left eye positions of the runner utilizing infrared light, and the third camera displayed the point of view of the runner as they traveled down a path. The gaze-tracker did not impair the vision of the runner, and the form of the runner was not impacted by the telemetry equipment worn on the participants’ backs. To calibrate the gaze-tracker, before running their trials participants looked at an ISCAN calibration board containing five points. At a distance of 3.0-m, participants first looked at the center point, followed by the top right, left and finally bottom left and right points consecutively. A distance calibration was also performed on the center calibration point to ensure calibration at different distances from the board. This resulted in an accurate projection onto the video frames of where the participants were looking. This projection was represented as a crosshair in each video frame (Fig 3).

The crosshair represents where the participants were looking in their visual field during the run.
Fig 3
This was digitized throughout each of the participants trials. The location of the digitized crosshair was used to determine if the participants were looking near or on the curb.The crosshair represents where the participants were looking in their visual field during the run.

Inclusion criteria for final analysis was based on several factors. In order to have a balanced sample to capture variation we selected six trials for each subject. We reduced the effect of novelty and learning during each run by choosing trials, across the entire range of the run for each participant (two at the beginning, two in the middle, and two at the end). To ensure quality in our data, the crosshair was recorded in the participants’ field of view for at least 84% of a trial. This was meant to reduce crosshair dropouts secondary to micromovements of the headset, interruption in the infrared camera secondary to ambient light exposure, as well as loss due to participant’s eyes looking beyond the peripheral limits of the tracking system (i.e., it was outside the view of the two small cameras positioned on the participant’s eyes). The majority of trials met this criterion. If one of the six trials selected had a cursor in fewer than 84% of the frames it was not used and another one from the same part of the trial set was selected. Therefore, a total of 60 trials with the cursor present in greater than 84% of the frames were analyzed (six per subject).

Objects in videos were digitized utilizing DLTv5 [41]. The crosshair (at its center) and curb (the four corners of the curb creating a quadrilateral) were digitized for the entire length of each trial, from the beginning of the running trial, to the time the curb disappeared from the participant’s vision at the end of the trial (Fig 3). All trials were examined for fixations. Those trials where the subject never fixated on the curb at any point were noted and included in analysis of the frequency of trials with no fixations (indicating on-line sampling throughout) and included in the analysis of fixation counts in bins, but not the timing of last fixation.

Algorithm

In studies of gaze it is a challenge to measure a fixation in a 2D projection of a visual field, i.e., to determine how close to the object the cursor must be to be scored as looking at the object. To resolve this issue, we designed an algorithmic automated system that incorporates system-based error and human-based (45 raters) judgments of what counted as attention to the curb. This was a multi-step process. First, to create a reasonable estimate of the area in the frame that would represent if the runner had placed their vision on the curb, the original digitized space was expanded slightly to include the error associated with the gaze tracker and potential calibration error (resulting in a quadrilateral that was slightly larger than the size of the curb itself). The amount of calibration error reported in the literature for the system used in this experiment and similar gaze tracking systems varies depending on the model and context in which it was used [11,27,33,42,43] (0.50°-2.00°). To determine an appropriate calibration error to be added around the digitized curb, an experiment was completed where the digitized curb was expanded on all sides at increments of 0.25° in error magnitude starting at 0.50° and moving to 2.00° (Fig 4).

The digitized curb created from the algorithm based on magnitudes of error.
Fig 4
The inset shows an actual video frame with the digitized curb (white dotted line, which equals 0.00° (the curb), and the red solid line equal to 2.00° magnitude of error, which was the maximum magnitude of error of the gaze-tracker found in similar studies. These error magnitudes were used to make algorithms, which were then compared to human raters to determine which error magnitude most closely replicated the interpretations of human raters. To create each increase in error for the curb, the error magnitudes (commonly expressed in degrees), were converted to pixels, allowing error to be matched with the units used in DLTdv5. The conversion factor for this chart in the X direction was 1 degree: 8.62 pixels and in the Y direction was 1 degree: 9.23 pixels. The central rectangle represents the curb (0.00°). For each successive increase in error magnitude from the original digitized curb through the largest error magnitude of 2.00° we added 4.30 pixels in the X direction and 4.62 pixels in the Y direction. The 1.25° error magnitude, shown in purple, was found to be the most representative of the interpretations of the human raters.The digitized curb created from the algorithm based on magnitudes of error.

To better quantify and identify the different regions near the curb, the video frame encompassing the field of view of the participant, was converted from degrees (76° x 52°, length by width) to pixels utilizing the conversion factor of 1°: 8.62 pixels in the X-direction and 1°: 9.23 pixels in the Y-direction (equating to a total of 655 pixels x 480 pixels). If the crosshair fell within one of these incremental zones (curb plus pixel, error of magnitude), then the algorithm indicated the runner looked at the curb (a “hit”) and if the crosshair was absent then the algorithm indicated the runner was not looking at the curb (a “miss”). This created an absence/presence tally across the running trials.

Then, to further automate the algorithm and evaluate it against human raters, 45 human raters examined twenty frames that had been evaluated by each algorithm (accounting for varying amounts of error) and rated each frame as a “hit” and a “miss.” Variability was noted in some of the frames (some raters determined that a frame was a “hit” while others determined it was a “miss”) and so a 65% threshold was determined to distinguish a hit from a miss. The threshold was calculated by using a Z-test and setting a 95% confidence interval on the hit/miss results from the group of raters (n = 45). Although more conservative than other choices, our algorithm determined “hits” that we were certain our raters felt represented the runner was looking at the curb. This allowed for comparison between the human rater’s judgment, often the standard in studies like this, and the designations given by the incremental zones of error magnitude within the algorithm. This two-part approach allowed for an algorithm to be chosen that most closely represented the perception of these human raters. During this comparison, when the raters called the frame a “hit” but the algorithm indicated it as a “miss,” this was defined as a “false miss (FM).” And when the raters called a frame a “miss” but the algorithm indicated it as a “hit” this was called a “false hit (FH).”

The third step in this process was to evaluate the concordance of each algorithm with human judgment. When an algorithm produced a large number of FMs, it was considered too conservative and when an algorithm produced a large number of FHs, it was considered too liberal. To measure reliability between raters’ decisions and computer decisions a Cohen’s Kappa (CK) [44] and a Goodman and Kruskal’s lambda [45] were both utilized. Both methods measure the agreement between each set of data and accounts for any agreement that might arise due to chance. A value for either statistic closer to 1.0, signifies higher levels of agreement between the raters and each algorithm. Importantly, a value of 0.0, indicates the two data sets are independent from one another. The difference between the two is that the Goodman and Kruskal’s lambda is somewhat more conservative than Cohen’s Kappa. With a Cohen’s Kappa of 0.89 and a Goodman and Kruskal’s lambda of 0.85, it was determined 1.25° of calibration error magnitude produced results most similar to the human raters. This error magnitude was then used on the remainder of our participants to determine if the participant was looking at the curb (Table 2; Fig 3). The error magnitude calculated for our study is similar to the magnitude of error found in Matthis and colleagues (2018), which is the most similar to our experimental design in that it was studying human locomotion in a natural environment utilizing a gaze-tracker.

table-wrap
Table 2
Summary of each Algorithm’s “Hit/Miss” Interpretations and Cohen’s Kappa measuring the reliability between the Algorithm and the Rater.
Hit ≥% Calibration Error (°) 0.50 0.75 1.00 1.25 1.50 1.75 2.00
65 False Miss 3 1 0 0 0 0 0
False Hit 0 0 0 0 2 2 2
Cohen’s Kappa 0.53 0.78 0.89 0.89 0.70 0.53 0.53

Data analysis

This definition of a fixation follows previous studies [2, 38,39,40], and in our study was determined as the point at which the runner’s gaze fell within the curb area (with 1.25° of calibration error applied), for at least three consecutive frames (defined as 99 ms). Therefore, a fixation in this study is taken to represent intentionality in the gaze of the runner and thus exhibited a period in which the runner was obtaining relevant information about their environment.

Sex-specific differences in the number of fixations per trial were assessed with a Mann-Whitney-U test. To better investigate fixations across the length of the running trials, each participant’s trials were divided into fifths based on the total time of the trials (each time increment was identified as a “Bin” yielding Bins 1, 2, 3, 4, and 5), which allowed for a summed average of fixations between individuals at different portions of their run. Bin 1 and 2 represent the early time increment, Bin 3 represents the middle time increment, and Bin 4 and 5 represent the end time increment (which ended when the curb was no longer in the field of view of the gaze tracker). Goodness of fit tests were utilized to compare the summed counts of fixations in each bin.

A series of Spearman’s rank correlation analyses were used to assess the relationship between runner speed and total number of fixations on the obstacle, and the relationship between runner speed and the timing of the last fixation. Timing was measured as a percent of the trial in which the last fixation occurred (meaning that for each trial with a fixation, the time period the last fixation occurred was normalized to the total time of the run and compared with all of the other trials). All statistical tests were conducted in PRISM (Graphpad Software, San Diego, CA).

Results

Total fixations, fixations across each trial, and timing of the participant’s last fixation

Fixations were observed in 80% (n = 48 out of 60) of trials. No subject had a complete set of trials with no fixation; that is all subjects fixated on the curb during some of their trials. Trials without fixations were fairly evenly distributed across subjects with half of the subjects fixating in every trial, one subject with one trial with no fixations, one subject with two trials with no fixations, and three subjects with three trials with no fixations. Thus, a total of 12 trials out of 60 appeared to use exclusively online sampling.

On average, participants fixated on the curb 2.4 ± 1.9 times during a trial (Fig 4). The number of fixations during a trial did not vary significantly (Mann-Whitney U = 11, p = 0.79) between males (2.4 ± 1.9) and females (2.3 ± 1.3). No significant difference was found when the summed counts were compared across bins using a Chi-Square Goodness of Fit (χ = 4.818; p = 0.31; Fig 5).

Summed average counts of fixations across participants split into bins.
Fig 5
No significant difference from the expected frequency noted in each bin utilizing a chi-square goodness of fit test.Summed average counts of fixations across participants split into bins.

Timing of last fixation for each trial

On average, the timing of the last fixation occurred at 59.2% ± 24.0% of the total trial time (Fig 4). No significant differences (Mann-Whitney U = 11, p = 0.84) were observed between the percent of trial in which the fixation last occurred for males and females (males = 62.9 ± 18.1% versus females = 55.5 ± 30.6%; P = 0.8413). However, the 60% average is driven in part by a bimodal distribution with half of the subjects showing the average last fixation occurring at approximately 40% of the run (roughly six meters from the curb) and the other half of the subjects average last fixation occurring at approximately 80% of the run (roughly 3 meters from the curb). It should be noted that one subject fixated intensely only at the end of the run in all their trials and may have driven the high values for the 80% average group.

Correlation analyses

Results of the Spearman’s rank correlation estimates demonstrated that there is no significant relationship between running speed and the timing of the last fixation (R = -0.02, P = 0.97). There was, however, a significant negative relationship between running speed and the average number of fixations within a trial (r = -0.76; P = 0.01; Fig 6).

The relationship between the average number of fixations for each participant and their average speed.
Fig 6
There is a strong negative relationship between how fast a participant was running and the number of average fixations (r = -0.76, P = 0.01).The relationship between the average number of fixations for each participant and their average speed.

Discussion

This study developed and tested hypotheses, based on previous laboratory and athletic field studies, about how human runners interact visually with a common obstacle (a typical sidewalk curb). The goal was to explore how much visual attention runners gave to a relatively common and easy to navigate obstacle, as opposed to a sudden or complex obstacle. To do this we recorded and quantified the fixations of runner’s as they approached the curb and identified where these fixations took place along the length of the run. Out of the 60 trials observed from our participants, 80% of the trials had at least one fixation on the curb, every participant had a trial with at least one fixation, and, on average, runners fixated on the cub 2.4 ± 1.5 times during each run. This rejects hypothesis H1o that subjects would not fixate on the curb at all. Online-sampling (as evidenced by the absence of fixation) is not a primary mechanism of attention at least during the majority of the duration of those runs.

Subjects had fixations throughout most of the trial and there were no differences in average fixation numbers between female and male participants. These results suggest that fixating on a region of interest in one’s path or environment at least once or twice during the approach phase is a common strategy among our participants and is important for implementing clearance maneuvers, proactive adjustments of body position, and route planning during running. Thus, the feedforward method of attention is seen here as a central part of obstacle management during running in a natural urban setting.

The results presented herein suggest two things about the latter portion of a run. First, on average the highest number of summed counts of fixations occurs between 60% and 80% of the run (Fig 4), a pattern consistent with previous studies of walking subjects [1, 2, 5, 9,10, 13,17, 23, 26, 31]. Second, this study found that individuals' average last fixation on the curb occurred in the last 20% - 40% of their run. There was no correlation between the runners’ final fixation time and their running speed, indicating that regardless of how fast a runner is moving, they still fixate near the middle of their run to judge the distance to an obstacle, and then proceed to clear an obstacle without the need for further fixations. The timing of one’s final fixation is also consistent with the feedforward model suggested by previous studies in walkers [1, 2, 5,10, 13,17], as it suggests that once runners are over halfway through their run, they have already pre-planned their path, adjusted their gait for the oncoming obstacle, and are looking onward for the next obstacle in their path. Further, it could be suggested that at this time in the run it may be advantageous for runners to transition from the more “proactive” feedforward aspect of vision to the more “improvisational” online control method, allowing runners to make quick on-the-fly adjustments to their gait to ensure safe navigation of the obstacle, while also planning for the next oncoming obstacle [6, 9, 11].

A negative correlation was found between the running speed of an individual and the total number of fixations within a trial. This suggests that faster runners who approach the curb relatively quickly have less time for feedforward adjustments and rely more heavily on the online control method in order to “fine-tune” their movements as they get closer to the obstacle [6, 9, 11].

There are several limitations to the current study. The relatively small sample size of 10 participants, all of whom were young and associated with Duke University may not be representative of the general running population. Including older runners, as well as those with different levels of running experience would be an important area for further research. However, these comparisons are beyond the scope of this study. In addition, when possible, we encourage future researchers to use more than 20 different gaze images to develop the criteria for their algorithm, as this sample size could have caused bias in our kappa and lambda statistics. It is also the case that our gaze-tracker utilizes infrared reflection lenses and that the use of this system during a running task is prone to movement of these lenses and thus could have disrupted our previous calibration.

To further understand the role of vision during obstacle clearance it would be worth examining saccades (i.e., shorter bouts of visual attention) as individuals cleared the obstacle [11]. In addition, studying the role of the runner’s peripheral vision in future studies is warranted to better characterize the online control method and the feedforward method which appear to predominate during clearance and approach, respectively. It would also be valuable to measure the step length of runners throughout the trial [46] and further, to examine if those that fixate on the curb more at the end of the trial change step length similarly to those that do not look at the curb as frequently. Additionally, using obstacles of different height similar to previous studies in walkers [26, 36] would be an interesting addition to a future study. Recently, Lucas-Cuevas et al. [47] demonstrated that, during treadmill running, the further that one’s gaze is from a region of interest, the worse one’s gait mechanics became while running and the less comfort runners had during that part of the experiment. It would be productive to look into the distance runners tend to focus outside of a ROI during safe obstacle navigation to get a sense as to how far from an obstacle individuals can focus and still obtain enough information to effectively and safely move down a path with a perturbation.

This study supports the hypothesis that runners will use a combination of feedforward and online sampling methods to navigate obstacles in a natural urban setting. For subjects in this study, visual attentiveness is part of a feedforward strategy for runners to plan and safely navigate an obstacle during the first 40% - 80% of the run, after which the runner shifts to an online-sampling method in which the curb is not a primary visual focus. Overall, this study adds to our understanding of the role of visual attention in human runners and is essential for understanding broader aspects of locomotor control and decision-making for runners as they navigate outdoor challenges.

Acknowledgements

The authors thank Angel Zeininger, Aidan Fitzsimons, and Megan Snyder for their contributions to the conception and design of this work.

References

Patla AE . Understanding the roles of vision in the control of human locomotion. Gait & Posture 1997; 5: 5469.

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 doi: 10.1097/00001756-199712010-00002

Land M , Mennie N , Rusted J . The Roles of Vision and Eye Movements in the Control of Activities of Daily Living. Perception 1999; 28: 13111328. doi: 10.1068/p2935

Fajen BR , Warren WH . Behavioral dynamics of steering, obstable avoidance, and route selection. Journal of Experimental Psychology: Human Perception and Performance 2003; 29: 343362. doi: 10.1037/0096-1523.29.2.343

Patla Aftab E. , and Vickers Joan N. How Far Ahead Do We Look When Required to Step on Specific Locations in the Travel Path during Locomotion?” Experimental Brain Research, 2003; 148: 13338. doi: 10.1007/s00221-002-1246-y

Mohagheghi AA , Moraes R , Patla AE . The effects of distant and on-line visual information on the control of approach phase and step over an obstacle during locomotion. Exp Brain Res 2004; 155: 459468. doi: 10.1007/s00221-003-1751-7

Reynolds RF , Day BL . Visual guidance of the human foot during a step. The Journal of Physiology 2005; 569: 677684. doi: 10.1113/jphysiol.2005.095869

Land MF . Eye movements and the control of actions in everyday life. Prog Retin Eye Res 2006; 25: 296324. doi: 10.1016/j.preteyeres.2006.01.002

Patla AE , Greig M . Any way you look at it, successful obstacle negotiation needs visually guided on-line foot placement regulation during the approach phase. Neuroscience Letters 2006; 397: 110114. doi: 10.1016/j.neulet.2005.12.016

10 

Rietdyk S , Rhea CK . Control of adaptive locomotion: effect of visual obstruction and visual cues in the environment. Exp Brain Res 2006; 169: 272278. doi: 10.1007/s00221-005-0345-y

11 

Marigold DS , Patla AE . Gaze fixation patterns for negotiating complex ground terrain. Neuroscience 2007; 144: 302313. doi: 10.1016/j.neuroscience.2006.09.006

12 

Shepherd SV , Platt ML . Spontaneous social orienting and gaze following in ringtailed lemurs. Anim Cogn 2008; 11: 13 doi: 10.1007/s10071-007-0083-6

13 

Matthis JS , Barton SL , Fajen BR . The biomechanics of walking shape the use of visual information during locomotion over complex terrain. Journal of Vision 2015; 15: 10–10.

14 

Drew T , Jiang W , Kably B , Lavoie S . Role of the motor cortex in the control of visually triggered gait modifications. Can J Physiol Pharmacol 1996; 74: 426442.

15 

Hollands MA , Marple-Horvat DE . Visually guided stepping under conditions of step cycle-related denial of visual information. Experimental Brain Research 1996; 109.

16 

Patla AE , Adkin A , Martin C , Holden R , Prentice S . Characteristics of voluntary visual sampling of the environment for safe locomotion over different terrains. Exp Brain Res 1996; 112: 513522. doi: 10.1007/BF00227957

17 

Whitwell RL , Milner AD , Cavina-Pratesi C , Barat M , Goodale MA . Patient DF’s visual brain in action: Visual feedforward control in visual form agnosia. Vision Research 2015; 110: 265276. doi: 10.1016/j.visres.2014.08.016

18 

Thomson JA . How do we use visual information to control locomotion? Trends in Neurosciences 1980; 3: 247250.

19 

Weerdesteyn V , Schillings AM , Galen GPV , Duysens J . Distraction Affects the Performance of Obstacle Avoidance During Walking. Journal of Motor Behavior 2003; 35: 5363. doi: 10.1080/00222890309602121

20 

Cho H , Romine NL , Barbieri FA , Rietdyk S . Gaze diversion affects cognitive and motor performance in young adults when stepping over obstacles. Gait & Posture 2019; 73: 273278.

21 

Lee DN , Lishman RJ , Thomson JA . Regulation of Gait in Long Jumping. Journal of Experimental Psychology: Human Perception and Performance 1982; 8: 44859.

22 

Bradshaw EJ , Sparrow W a . The Speed-Accuracy Trade-Off in Human Gait Control When Running Towards Targets. Journal of Applied Biomechanics 2000; 16: 331341.

23 

Bradshaw EJ , Sparrow WA . Effects of approach velocity and foot-target characteristics on the visual regulation of step length. Human Movement Science 2001: 26.

24 

Weerdesteyn V , Nienhuis B , Hampsink B , Duysens J . Gait adjustments in response to an obstacle are faster than voluntary reactions. Human Movement Science 2004; 23: 351363. doi: 10.1016/j.humov.2004.08.011

25 

Weerdesteyn V , Nienhuis B , Duysens J . Advancing age progressively affects obstacle avoidance skills in the elderly. Human Movement Science 2005; 24: 865880. doi: 10.1016/j.humov.2005.10.013

26 

Bradshaw EJ , Aisbett B . Visual guidance during competition performance and run‐through training in long jumping. Sports Biomechanics 2006; 5: 114. doi: 10.1080/14763141.2006.9628221

27 

Marigold DS , Weerdesteyn V , Patla AE , Duysens J . Keep looking ahead? Re-direction of visual fixation does not always occur during an unpredictable obstacle avoidance task. Exp Brain Res 2007; 176: 3242. doi: 10.1007/s00221-006-0598-0

28 

Samir Matthis Jonathan , Fajen Brett R . Humans exploit the biomechanics of bipedal gait during visually guided walking over complex terrain. Proceedings of the Royal Society B: Biological Sciences 2013; 280.

29 

Matthis JS , Fajen BR . Visual control of foot placement when walking over complex terrain. Journal of Experimental Psychology: Human Perception and Performance 2014; 40: 106115. doi: 10.1037/a0033101

30 

Matthis JS , Barton SL , Fajen BR . The biomechanics of walking shape the use of visual information during locomotion over complex terrain. Journal of Vision 2015; 15: 1010.

31 

Matthis JS , Barton SL , Fajen BR . The critical phase for visual control of human walking over complex terrain. Proceedings of the National Academy of Sciences 2017; 114: E6720E6729.

32 

Stuart S , Hunt D , Nell J , Godfrey A , Hausdorff JM , Rochester L et al. Do you see what I see? Mobile eye-tracker contextual analysis and inter-rater reliability. Med Biol Eng Comput 2017; 56: 289296. doi: 10.1007/s11517-017-1669-z

33 

Matthis JS , Yates JL , Hayhoe MM . Gaze and the Control of Foot Placement When Walking in Natural Terrain. Curr Biol 2018; 28: 12241233. doi: 10.1016/j.cub.2018.03.008

34 

Ryu D , Abernethy B , Mann DL , Poolton JM . “The contributions of central and peripheral vision to expertise in basketball: How blur helps to provide a clearer picture.” Journal of Experimental Psychology: Human Perception and Performance 2015; 41:16785. doi: 10.1037/a0038306

35 

Vater C , Williams AM , Hossner E-J . “What do we see out of the corner of our eye? The role of visual pivots and gaze anchors in sport.” International Review of Sport and Exercise Psychology 2019:123.

36 

Diaz Gabriel J. , et al The Pickup of Visual Information about Size and Location during Approach to an Obstacle.” PLOS ONE, 2018: 13.

37 

Bradshaw , Elizabeth J. , and Sparrow W. A . “Effects of Approach Velocity and Foot-Target Characteristics on the Visual Regulation of Step Length.” Human Movement Science, 2001: p. 26.

38 

Fowler GA , Sherk H . Gaze during visually-guided locomotion in cats. Behavioural Brain Research 2003; 139: 8396. doi: 10.1016/s0166-4328(02)00096-7

39 

Franchak JM , Adolph KE . Visually Guided Navigation: Head-Mounted Eye-Tracking of Natural Locomotion in Children and Adults. Vision Res 2010; 50: 27662774. doi: 10.1016/j.visres.2010.09.024

40 

Kano F , Tomonaga M . Head-Mounted Eye Tracking of a Chimpanzee under Naturalistic Conditions. PLOS ONE 2013; 8: e59785 doi: 10.1371/journal.pone.0059785

41 

Hendrick T. Software techniques for two- and three-dimensional kinematic measurements of biological and biomimetic systems—IOPscience. 2008.http://iopscience.iop.org/article/10.1088/1748-3182/3/3/034001/meta (accessed 27 Dec2017).

42 

Shepherd SV , Platt ML . Noninvasive telemetric gaze tracking in freely-moving socially-housed prosimian primates. Methods 2006; 38: 185194. doi: 10.1016/j.ymeth.2005.12.00

43 

Johnson JS , Liu L , Thomas G , Spencer JP . Calibration algorithm for eyetracking with unrestricted head movement. Behav Res 2007; 39: 123132.

44 

McHugh ML . Interrater reliability: the kappa statistic. Biochem Med (Zagreb) 2012; 22: 276282.

45 

Hartwig Frederick . “Statistical Significance of the Lambda Coefficients.” Behavioral Science, 1973; 18: 307310.

46 

Larsen RJ , Jackson WH , Schmitt D . Mechanisms for regulating step length while running towards and over an obstacle. Human Movement Science 2016; 49: 186195. doi: 10.1016/j.humov.2016.07.002

47 

Lucas-Cuevas ÁG , Priego Quesada JI , Gooding J , Lewis MGC , Encarnación-Martínez A , Perez-Soriano P . The effect of visual focus on spatio-temporal and kinematic parameters of treadmill running. Gait & Posture 2018; 59: 292297.

16 Apr 2020

PONE-D-20-07008

Gaze-Behaviors of Runners in a Natural, Urban Running Environment

PLOS ONE

Dear Mr. Cullen,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

We would appreciate receiving your revised manuscript by May 31 2020 11:59PM. When you are ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter.

To enhance the reproducibility of your results, we recommend that if applicable you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

Please include the following items when submitting your revised manuscript:

    A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). This letter should be uploaded as separate file and labeled 'Response to Reviewers'.

    A marked-up copy of your manuscript that highlights changes made to the original version. This file should be uploaded as separate file and labeled 'Revised Manuscript with Track Changes'.

    An unmarked version of your revised paper without tracked changes. This file should be uploaded as separate file and labeled 'Manuscript'.

Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.

We look forward to receiving your revised manuscript.

Kind regards,

Nizam Uddin Ahamed, PhD

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

1. Please amend your manuscript to include your abstract after the title page.

2. Please include captions for your Supporting Information files at the end of your manuscript, and update any in-text citations to match accordingly. Please see our Supporting Information guidelines for more information: http://journals.plos.org/plosone/s/supporting-information.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Partly

Reviewer #2: Yes

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: No

Reviewer #2: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: The study attempts to answer the question of whether runners fixate on obstacles such as sidewalk curbs when running in an urban road setting. To answer the question, authors followed a dual-process visual control framework, used a binocular gaze-tracker, and developed a customized algorithm to test gazing behaviors among runners in two primary hypotheses. Results showed that runners tended to use a combination of online and feedforward sampling control strategies and that faster average running pace led to fewer fixations on the obstacle of sidewalk curb. In many ways the manuscript is interesting, particularly regarding the development of customized eye-tracking algorithm in dealing with visual fixations on objects in 3D space. It is also well written. With that said, I do have several concerns that could help enhance the manuscript. The concerns were elaborated below.

First, I have concerns about employing the current eye-tracking system in a running task. Based on what I can find given the information authors provided, the eye-tracking system seems to involve infrared (IR) reflection lenses (i.e., the hot mirrors) in tracking the eyes. Such a design of IR lenses, compared to those embedded in the glasses frames in some other eye-tracking brands, may be less fit to studying the running task. It is because running movements involve constant bodily acceleration that can dislocate previously calibrated IR reflection lenses, which is usually connected to the glasses frame via weak plastic connectors. I understand that, if I am correct about the hardware, the authors may have limited power in attenuating the above issue associated with the equipment design, but I encourage them to fully acknowledge such difficulties and include a picture of the exact eye-tracking system used.

Another challenge of tracking runners’ vision comes from the fact that runners generally need to keep their upper bodies (including heads) relatively stable during running to maximize movement efficiency. That is, keeping the head still during running requires more visual search to be achieved by replacing head movements with eyeball movements, which makes the pupils more likely to appear at off-center locations in the eyes. For the majority of eye-tracking systems utilizing the ‘pupil to CR (corneal reflection)’ technique, big eyeball movements would cause technical issues by making the IR dots more likely to fall off the pupil region and by reshaping the appearance of the pupil from a circle to a oval shape, which is harder for the algorithm to recognize. That is to say, the task would cause increased likelihood for tracking loss. Although authors reported the inclusion criterion of 84% (l.223), I worry about potential dependence between tracking loss and gazing behaviors in the running task, which can result in bias regarding evidence analyzed and conclusions. I would encourage authors to do some analyses comparing those low-tracking rate trials (e.g., the lowest 6 trials) with the current evidence analyzed so that we can either eliminate or confirm such a possible bias, either way it would be a contribution.

Furthermore, although the authors did a good job reviewing literature of foveal vision in relevant task models (to some extent, I feel that the authors overly cited the literature in several locations in the Introduction, so I suggest to cut off some redundant ones), the literature on peripheral vision is largely missing. It makes sense that the foveal vision should be a focus of literature review given the use of eye-tracking, which is really foveal vision tracking. However, visual attention can come from or even be dominated by peripheral vision. It is likely that in some instances runners rely more on peripheral vision in monitoring environments. I would recommend authors give a light weighted review (and also discussion) on peripheral vision. A recent review from Vater et al. would be a good start place.

Vater, C., Williams, A. M., & Hossner, E. J. (2019). What do we see out of the corner of our eye? The role of visual pivots and gaze anchors in sport. International Review of Sport and Exercise Psychology, 1-23.

Since the algorithm is supposed to be a major contribution of the manuscript, I have some suggestions to enhance it: 1. try to include Goodman and Kruskal’s Lambda in addition to Cohen’s Kappa. Lambda could give another view of the ‘rater agreement’; 2. explicitly acknowledge the small number of frames (i.e., 20, l. 279) used in testing the algorithm and encourage future studies to increase it. Such a small number may generate biased Kappa and Lambda estimates and is subjected to selection risks (i.e., why this 20 out of potentially hundreds of frames? why not randomly draw from the pool of available frames?).

Lastly, I have some comments on the results section. Authors non-parametrically tested a small sample (n=10). None of the results in Section 3.1 and 3.2 reached significance (p < .05), which is not surprising given the compromised statistical power from both directions. The only significant results came from a parametric test (e.g., simple regression) in Section 3.3. To be consistent in authors’ testing approach, shouldn’t they replace the a simple regression (which basically gave estimates on Pearson r) with Spearman Rank correlation estimates?

Reviewer #2: The study tackled an interesting topic on gaze behavior of the runners in facing the natural obstacles in terms of pre-planning vs. online planning which is important to the respective field, the statistical analyses and results were sound and the manuscript was well-written.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #2: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files to be viewed.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at figures@plos.org. Please note that Supporting Information files do not need this step.


24 Apr 2020

We want to thank the editor and the two anonymous reviewers for comments and suggestions. We were able to accommodate every suggestion and we think that the changes we made have improved the manuscript. We are grateful for the opportunity to submit a revised version and hope that you find this manuscript acceptable for publication.

Reviewer #1: The study attempts to answer the question of whether runners fixate on obstacles such as sidewalk curbs when running in an urban road setting. To answer the question, authors followed a dual-process visual control framework, used a binocular gaze-tracker, and developed a customized algorithm to test gazing behaviors among runners in two primary hypotheses. Results showed that runners tended to use a combination of online and feedforward sampling control strategies and that faster average running pace led to fewer fixations on the obstacle of sidewalk curb. In many ways the manuscript is interesting, particularly regarding the development of customized eye-tracking algorithm in dealing with visual fixations on objects in 3D space. It is also well written. With that said, I do have several concerns that could help enhance the manuscript. The concerns were elaborated below.

We want to thank the reviewer for their compliments and the suggestions for ways to improve the paper. We have been able to take all these suggestions into account and believe that the paper is significantly improved.

First, I have concerns about employing the current eye-tracking system in a running task. Based on what I can find given the information authors provided, the eye-tracking system seems to involve infrared (IR) reflection lenses (i.e., the hot mirrors) in tracking the eyes. Such a design of IR lenses, compared to those embedded in the glasses frames in some other eye-tracking brands, may be less fit to studying the running task. It is because running movements involve constant bodily acceleration that can dislocate previously calibrated IR reflection lenses, which is usually connected to the glasses frame via weak plastic connectors. I understand that, if I am correct about the hardware, the authors may have limited power in attenuating the above issue associated with the equipment design, but I encourage them to fully acknowledge such difficulties and include a picture of the exact eye-tracking system used.

Thank you for your comment. This was a major concern for us as well. Yet we really wanted to expand the use of these systems into dynamic settings and we hope we made adjustments that make this system work well. The current system is the most modern designed by iScan and we worked with them to make it as sturdy as possible for this goal when we purchased. In addition, we designed the helmet with the tight fitting band on the head that limited movement and firmly affixed the glasses in places (without discomfort). We have inserted a new figure, Figure 2 in the manuscript, of one of our authors (MC) wearing our gaze-tracker, helmet and backpack. We did all our runs at dusk and dawn to minimize outdoor light exposure. We have expanded on that in the methods (starting on line 194) and have acknowledged these limitations in our methods and discussion (line 458-461). In addition, recognizing the limits we considered larger regions of gaze (the curb broadly speaking and with error incorporated) to avoid some of those issues. We hope the reader will consider these limitations and that future studies will further address design.

Another challenge of tracking runners’ vision comes from the fact that runners generally need to keep their upper bodies (including heads) relatively stable during running to maximize movement efficiency. That is, keeping the head still during running requires more visual search to be achieved by replacing head movements with eyeball movements, which makes the pupils more likely to appear at off-center locations in the eyes. For the majority of eye-tracking systems utilizing the ‘pupil to CR (corneal reflection)’ technique, big eyeball movements would cause technical issues by making the IR dots more likely to fall off the pupil region and by reshaping the appearance of the pupil from a circle to a oval shape, which is harder for the algorithm to recognize. That is to say, the task would cause increased likelihood for tracking loss. Although authors reported the inclusion criterion of 84% (l.223), I worry about potential dependence between tracking loss and gazing behaviors in the running task, which can result in bias regarding evidence analyzed and conclusions. I would encourage authors to do some analyses comparing those low-tracking rate trials (e.g., the lowest 6 trials) with the current evidence analyzed so that we can either eliminate or confirm such a possible bias, either way it would be a contribution.

Thank you for your comment and we feel that this is a great point. We went through all of our trials that were excluded either because they were not at the correct portion of the run (beginning, middle, or end) or did not meet our 84% inclusion criteria. We found that 49% of our rejected trials met the 84% or greater inclusion criteria. As a result both low and high cursor trials were excluded. When we expanded this to 80%, it included 71% of our sample, and finally when expanded to 75%, this included 96% of our sample. We hope that this data shows that the trials that were excluded had the crosshair present for a similar amount of time as the trials we included and that often it was the time of the run (beginning, middle, and end) that determined what trial was chosen rather than a trial having the highest percentage of crosshair presence. The statement “For those with more than six, the trials with the six highest percentages of crosshair presence were included,” has been removed and the methods (lines 236- 247) have been modified to reflect that fact that trials were chosen from the beginning, middle, and end of the run, and that each of these met our 84% inclusion criteria. We would like to emphasize that this does not mean that the trials chosen for a particular individual included the highest percentage of crosshair presence, it was the fact that it met our criteria of having the crosshair for greater than 84% of the trial and the trial was at the correct portion of the run.

Furthermore, although the authors did a good job reviewing literature of foveal vision in relevant task models (to some extent, I feel that the authors overly cited the literature in several locations in the Introduction, so I suggest to cut off some redundant ones), the literature on peripheral vision is largely missing. It makes sense that the foveal vision should be a focus of literature review given the use of eye-tracking, which is really foveal vision tracking. However, visual attention can come from or even be dominated by peripheral vision. It is likely that in some instances runners rely more on peripheral vision in monitoring environments. I would recommend authors give a light weighted review (and also discussion) on peripheral vision. A recent review from Vater et al. would be a good start place.

Vater, C., Williams, A. M., & Hossner, E. J. (2019). What do we see out of the corner of our eye? The role of visual pivots and gaze anchors in sport. International Review of Sport and Exercise Psychology, 1-23.

Thank you for the reference this helped us really consider this issue. We have added references to the manuscript and have added to our introduction more on peripheral vision during the approach (lines 118-121) and have emphasized its importance during clearance of the obstacle (lines 121-122). We added a sentence on the need for further exploration of the study of peripheral vision in running to better characterize our understanding of the approach and navigation of obstacles as human’s run toward them in the discussion (line 465-467).

Since the algorithm is supposed to be a major contribution of the manuscript, I have some suggestions to enhance it: 1. try to include Goodman and Kruskal’s Lambda in addition to Cohen’s Kappa. Lambda could give another view of the ‘rater agreement’; 2. explicitly acknowledge the small number of frames (i.e., 20, l. 279) used in testing the algorithm and encourage future studies to increase it. Such a small number may generate biased Kappa and Lambda estimates and is subjected to selection risks (i.e., why this 20 out of potentially hundreds of frames? why not randomly draw from the pool of available frames?).

This is a great suggestion, thank you. We have calculated the Goodman and Kruskal’s lamba and have found that the pattern is fundamentally the same. We have included the value in the methods section with the Cohen’s Kappa (line 322). We agree that future studies should add more frames where possible. We would like to acknowledge that we had a limited number of frames with diversity (i.e. frames were meant to be a representative sample of some where the individual was clearly not or was looking at the curb, as well as a representative sample of some that were more ambiguous). We have acknowledged this limitation at the end of the in the manuscript and have a sentence encouraging others to increase the number of frames in the future (lines 456-458).

Lastly, I have some comments on the results section. Authors non-parametrically tested a small sample (n=10). None of the results in Section 3.1 and 3.2 reached significance (p < .05), which is not surprising given the compromised statistical power from both directions. The only significant results came from a parametric test (e.g., simple regression) in Section 3.3. To be consistent in authors’ testing approach, shouldn’t they replace the a simple regression (which basically gave estimates on Pearson r) with Spearman Rank correlation estimates?

Thank you for your comment. We have changed our parametric simple regression to Spearman Rank correlation estimates (Figure 6 and lines 395-398).

Reviewer #2: The study tackled an interesting topic on gaze behavior of the runners in facing the natural obstacles in terms of pre-planning vs. online planning which is important to the respective field, the statistical analyses and results were sound and the manuscript was well-written.


28 Apr 2020

PONE-D-20-07008R1

Gaze-Behaviors of Runners in a Natural, Urban Running Environment

PLOS ONE

Dear Mr. Cullen,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

We would appreciate receiving your revised manuscript by Jun 12 2020 11:59PM. When you are ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter.

To enhance the reproducibility of your results, we recommend that if applicable you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

Please include the following items when submitting your revised manuscript:

    A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). This letter should be uploaded as separate file and labeled 'Response to Reviewers'.

    A marked-up copy of your manuscript that highlights changes made to the original version. This file should be uploaded as separate file and labeled 'Revised Manuscript with Track Changes'.

    An unmarked version of your revised paper without tracked changes. This file should be uploaded as separate file and labeled 'Manuscript'.

Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.

We look forward to receiving your revised manuscript.

Kind regards,

Nizam Uddin Ahamed, PhD

Academic Editor

PLOS ONE

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: The revision of the manuscript reasonably address all the comments with one exception. Regarding the switch from using Pearson correlation to Spearman rank correlation, corresponding changes/edits shall be made at the section of ‘Data analysis’ (lines 339-344) and the figure illustrating the correlation between fixation number and running speed shall be updated to reflect the tested (i.e., rank) instead of original variable metrics metrics.

I would endorse the acceptance of this manuscript after seeing these changes, which is necessary for a coherently presented manuscript.

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: Sicong Liu

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files to be viewed.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at figures@plos.org. Please note that Supporting Information files do not need this step.


28 Apr 2020

We want to thank the editor, Dr. Liu, and our anonymous editor for their comments. We were able to accommodate your suggestions and we think that the changes have improved the manuscript. We are grateful for the opportunity to resubmit our manuscript for publication.

Response to Reviewers

Reviewer #1: The revision of the manuscript reasonably address all the comments with one exception. Regarding the switch from using Pearson correlation to Spearman rank correlation, corresponding changes/edits shall be made at the section of ‘Data analysis’ (lines 339-344) and the figure illustrating the correlation between fixation number and running speed shall be updated to reflect the tested (i.e., rank) instead of original variable metrics.

I would endorse the acceptance of this manuscript after seeing these changes, which is necessary for a coherently presented manuscript.

Thank you for your comment, we are sorry we did not address this point fully in our last revision. We have adjusted line 339 to state “Spearman’s rank correlation” rather than “ordinary least squares regression.” We have also removed the regression line from the Spearman’s rank correlation in Figure 6.


30 Apr 2020

Gaze-Behaviors of Runners in a Natural, Urban Running Environment

PONE-D-20-07008R2

Dear Dr. Cullen,

We are pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it complies with all outstanding technical requirements.

Within one week, you will receive an e-mail containing information on the amendments required prior to publication. When all required modifications have been addressed, you will receive a formal acceptance letter and your manuscript will proceed to our production department and be scheduled for publication.

Shortly after the formal acceptance letter is sent, an invoice for payment will follow. To ensure an efficient production and billing process, please log into Editorial Manager at https://www.editorialmanager.com/pone/, click the "Update My Information" link at the top of the page, and update your user information. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to enable them to help maximize its impact. If they will be preparing press materials for this manuscript, you must inform our press team as soon as possible and no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

With kind regards,

Nizam Uddin Ahamed, PhD

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: (No Response)

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: Sicong Liu


4 May 2020

PONE-D-20-07008R2

Gaze-Behaviors of Runners in a Natural, Urban Running Environment

Dear Dr. Cullen:

I am pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please notify them about your upcoming paper at this point, to enable them to help maximize its impact. If they will be preparing press materials for this manuscript, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

For any other questions or concerns, please email plosone@plos.org.

Thank you for submitting your work to PLOS ONE.

With kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Nizam Uddin Ahamed

Academic Editor

PLOS ONE

This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
https://www.researchpad.co/tools/openurl?pubtype=article&doi=10.1371/journal.pone.0233158&title=Gaze-behaviors of runners in a natural, urban running environment&author=&keyword=&subject=Research Article,Biology and Life Sciences,Physiology,Biological Locomotion,Running,Medicine and Health Sciences,Physiology,Biological Locomotion,Running,Biology and Life Sciences,Neuroscience,Sensory Perception,Vision,Biology and Life Sciences,Psychology,Sensory Perception,Vision,Social Sciences,Psychology,Sensory Perception,Vision,Biology and Life Sciences,Physiology,Biological Locomotion,Walking,Medicine and Health Sciences,Physiology,Biological Locomotion,Walking,Biology and Life Sciences,Neuroscience,Cognitive Science,Cognitive Psychology,Attention,Biology and Life Sciences,Psychology,Cognitive Psychology,Attention,Social Sciences,Psychology,Cognitive Psychology,Attention,Physical Sciences,Mathematics,Applied Mathematics,Algorithms,Research and Analysis Methods,Simulation and Modeling,Algorithms,Biology and Life Sciences,Physiology,Biological Locomotion,Medicine and Health Sciences,Physiology,Biological Locomotion,Biology and Life Sciences,Physiology,Sensory Physiology,Visual System,Medicine and Health Sciences,Physiology,Sensory Physiology,Visual System,Biology and Life Sciences,Neuroscience,Sensory Systems,Visual System,Biology and Life Sciences,Anatomy,Head,Eyes,Medicine and Health Sciences,Anatomy,Head,Eyes,Biology and Life Sciences,Anatomy,Ocular System,Eyes,Medicine and Health Sciences,Anatomy,Ocular System,Eyes,