Source of Acquisition NASA Johnson Space Center STRIDE-CYCLE INFLUENCES ON GOAL-DIRECTED HEAD MOVEMENTS MADE DURING WALKING Brian T. Richasd E.A. van Emmerik2, and Jacob J. Bloomberg3 Neuroscience Laboratoiies, Wyle Life Sciences, Houston, TX, USA Department of Exercise Science, University of Massachusetts, Amherst, MA, USA Neuroscience Laboratoiies, NASA/Johnson Space Center, Houston, TX, USA Subtitle: Head Movements Made While Walking Coil-espondence: Brian T. Peters Wyle Life Sciences 1290 Hercules Dr. Suite 120 Houston, TX 77058, USA Phone: (281) 244-6574; Fax: (281) 244-5734 e-mail: bpeters @ems.js c.nasa.gov Abstract Horizontal head movements were studied in six subjects as they made rapid hoi-izontal gaze adjustments while walking. The aim of the present research was to deteimine if gait-cycle events alter the head movement response to a visual target acquisition task. Gaze shifts of approximately 40" were elicited by a step change in the position of a visual target from a central location to a second location in the left or i-ight horizontal pei-ipheiy. The timing of the target position change was constrained to occur at 25,50,75 and 100% of the stride cycle. The trials were randomly presented as the subjects walked on a treadmill at their prefeiyed speed (range: 1.25 to 1.48 m/s, mean: 1.39 0.09 d s ) . Analyses focused on the movement onset latencies of the head and eyes 1: and on the peak velocity and saccade amplitude of the head movement response. A comparison of the group means indicated that the head movement onset lagged the eye onset (262 ms versus 252 ms). The head and eye movement onset latencies were not affected by either the direction of the target change nor the point in the gait cycle during which the target relocation occm-ed. However, the presence of an interaction between the gait cycle events and the direction of the visual target shift indicates that the peak head saccade velocity and head saccade amplitude are affected by the natural head oscillations that occur while walking. Key words: head movement, target acquisition, locomotion, eye-head coordination, vision Introduction Visual fixation of targets peripheral to the progression of travel is common as we walk, yet gaze adjustments made duiing locomotion have not been systematically studied. When these gaze adjustments are sufficiently large, a head rotation must accompany the eye movements to achieve the goal. Little is known about the interaction between these goal-directed head movements and the periodic head translations and rotations that naturally accompany locomotion. Repoi-ts vaiy regarding the magnitude of these peiiodic head movements that occur while walking, but the phase relationship between these movements and ongoing gait cycle is consistent'-8. In the sagittal plane for example, the head pitches upward as the body translates down during each step. Yaw head movements counter the lateral translations of the body in a similar way. Based on studies using passive rotations of the body, it is believed that these movements are the result largely of reflexive mechanismsgy Regardless of the control mechanism (Le. reflexive, voluntay, or passive inertial) lo. these head movements imply the presence of modulating levels of neck muscle activation and vestibular stimulation. The effects of this modulation on goal-dliected head movements are unknown. Bent and colleagues found differences in lower body gait parameters that were dependent on when in the gait cycle a vestibular stimulus was given12. A head saccade made while walking could impart a similar vestibular disturbance. The results of a study comparing the effects of voluntaiy and unexpected head turns while walking, led Vallis and Patla to conclude that the central nervous system partially nullifies the sensoiy input created by voluntary head tui~~sI'n~ a.d dition to this proposed use of efferent copy to minimize the vestibular disturbance, a strategy of triggei-ing head movements to coincide with phases of the gait cycle where the vestibular infoimation is less impoi-tant could also be used. Such a strategy would affect the timing of the head movement response. Differences in the response timing may also result from interactions with other ongoing activities. Lajoie et al. showed that reaction time was dependent on when in the gait-cycle the stimulus was presented". The authors attributed this result to varying levels of attentional demand across the gait cycle. 2 In addition to possible changes in the initiation of a response, the process of superimposing a voluntslly head movement on the naturally occurring motions may affect the dynamics of the head movement response. The eye-head coordination during gaze re-fixations has .been studied extensively in seated subjects, but this behavior has not been studied in waking subjects. Changes in the head movement dynamics could impose limitations on the total response time (i.e. reaction time plus movement time) and could result in final head positions that leave the eyes at orbital eccenhicities that are sub-optimal for visual acuity. Through measures of movement onset latency and head movement kinematics, the goal of the present research was to assess the effects of gait cycle events on head movements made to acquire targets in the visual pei-ipheiy. Materials and Methods Subjects Six healthy males in the age range from 26 to 35 (mean 3 1 years 24) served as subjects for this study. None of the subjects had complaints of neck soreness at the time of the test and none had any history of vestibular disease. The expeiiment protocol was approved by the University of Massachusetts’ Human Subjects Review Committee and all subjects gave their informed consent piior to participation. Testing Conditions Walking Speed The subjects, all wearing a similar type jogging shoe, were tested as they walked on a motor- driven h-eadrnill (Accud P, Pacer Fitness Systems, Inc., Irving, TX) at their prefei-red walking speed The subject-specific speeds were established using an interactive method defined in Molt et al l4 which resulted in speeds ranging fiom 1.25 to 1.48 m/s (mean 1.39 -c 0.09 d s ) . Visual Targets Visual targets were presented on a rear-projection screen @a-Lite Screen Company, Inc., Warsaw, Indiana) that was placed between the subject and an LCD projector (Sharp, Model: XG- E670U, Mahwah, New Jersey). The screen was perpendicular to the walking direction at a distance 3 of -1 10 cm from the subject’s nasal bridge. The targets, consisting of alpha-numeiic characters, were controlled by custom software (LabVIEW; National Instruments Coip., Austin, TX.). When presented directly in front of the subject, the visual targets subtended a visual angle of -1.3”. During each of the 14-second walking trials, the visual target was initially presented in the center of the subject’sf ield-of-view at a height that was perceived by them to be eye level. The presented targets were randomly selected from a “pool” that contained 24 upper-case characters (0 and I were excluded) and 8 numerals (0 and 1 were omitted). With the exception of the first target to appear in the lateral position, which was visible for 900 ms, the display duration for the sequentially displayed characters was assigned randomly using the finite set of times that fell at 50 ms intei-vals from 400 to 700 ms. Targets in the lateral positions were restricted to the pool of 24 letters and presented at the same height as the central target but at positions that were 1 m to the left or iight of it. A gaze compensation angle of -r40” was required to visually fixate the lateral targets that continued to be displayed until the end of the trial. At the conclusion of each trial the subject ‘s recollection of the number of numerals that appeared in the central position, as well as the first letter seen in the lateral position, was compared to the actual values. While this information was used to establish an inclusion criterion for acceptable data trials, the piimary intent of requiring this from the subject was to motivate them to concentrate on’the task. The timing of the target change from the central position to the lateral position was one of the ciitical aspects of this investigation. The stimulus-generating program used a signal produced by a roller switch secured to the lateral edge of the heel of the iight shoe to control this change. Measured from the right heel strike, the target position changes were restricted to four phases, or quadrants, coi-responding to 25,50,75 and 100% of the stride cycle. The two lateral positions and four stride quadrants created eight possible test conditions, which were each repeated three times per subject. The order of the twenty-four tiials was randomized piior to the start of data collection. 4 Standing test Although a comparison between walking and standing conditions was not the intended focus of this investigation, relatively long reaction time latencies during the walking task prompted us to repeat the task during a subsequent test session. Five of the six subjects repeated the gaze re- fsation task as they stood on the non-moving treadmill belt. Measurement Sy st ems Kinematic and analog variables were collected at 240 Hz using the Qualisys motion analysis system (Qualisys, Inc., Glastonbm-y, CT.). The system monitored the positions of seven retro- reflective markers affixed to a light-weight headband (183 g) and a torso harness. The relative position of a virtual marker placed on the subject's nasal bridge was determined piioi- to the start of the data collection. This infoimation allowed the location of that landmark to be calculated throughout each of the data trials. The analog signals consisted of two TTL outputs that were used to indicate light heel strike and the timing of visual target position change and hoiizontal eye movements. The eye movements were obtained using standard DC Electrooculography (EOG) techniques. Data Processing Head Yaw Calculation A line segment connecting the virtual nasal biidge marker and the marker placed on the occipital region of the head was used to calculate the yaw movements of the head. Because this line segment deviated only slightly above and below (- 1-2") the transverse plane while waking any angle calculation errors introduced by out-of-plane motions were minimal. Movement Onset Detei-mination The onset of the saccadic movements for both the head and eyes was considered to be the point at which their velocity trajectoiies deviated from a 21 SD band around the average per-stride trajectoiy. A fourth-order, zero-lag Butteiwoi-th filter was applied to the raw position data piior to 5 calculating the velocities with a 3-point central difference ggoiithm. The per-stride average trajectories and standard deviations (SD) were calculated for each hid using the first six complefe strides, each of which had been linearly interpolated to 250 points. Because the 5 Hz filter applied to the data prior to determining the average trajectory significantly attenuated the shai-pness of the saccadic movement, the epoch containing the saccade was filtered using 15 Hz cut-off. The point at which this signal deviated from the average trajectoiy was automatically deteimined using custom- wiitten software (MATLAB; The Mathworks, Inc., Natick, MA). An operator who was sufficiently blinded from knowing the condition of the trial being displayed, visually inspected each trial and was able to manually input the appropriate position if the location of the automatically determined movement onsets was incoirect. Peak Yaw Head Velocity The peak yaw head velocity was deteimined to be the maximum value observed in the filtered (15 Hz cutoff) yaw velocity. This value was also visually inspected from the time series data. Yaw Head Saccade Amditude Variability in the stride-to-stride trajectory of the yaw head position prevented the "offset" of the saccade from being reliably determined. In lieu of using the offset to deteimine the amplitude of the head saccade, it was calculated to be the difference between the average yaw position of the head during the first complete stride after the saccade and the average position of the last complete stride before the saccade. Statistical Analysis Piior to perfoiming any statistical analyses, data from 5 of the 120 trials were excluded from the data set because the target position change occurred outside a 55% window of the desired stride position (i.e. 20 - 30% for the 25% condition). Another 19 trials were eliminated because the subjects did not accurately identify the first letter that appea-ed in the lateral position. This was usually the result of a response that was sufficiently delayed due to either starting the movement in the opposite direction or because of a lack of attention. An ANOVA using a split-plot factorial 6 design was used for the statistical analyses. The ANOVA model contained tests for main effect differences between SUBJECTS, TRIALS ,t arget step DIRECTION, 2nd gait-cycle QUADRANT. It also included a test for a DIRECTION x QUADRANT interaction that was nested within the SUBJECTS. The test for TRIALS effects was included to make the analysis complete, but as would be expected given the randomized presentation of the 24 trials this comparison showed no effect. Therefore, it is not included in the discussion of the results. Results Basic Head and Eye Motions in Yaw While walking and attending to the visual target, the yaw head motion of all of the subjects had an oscillatory nature that repeats itself with each stride (i.e. heelsti-ike to heelsti-ike of the same foot). This oscillation is present during gaze fixation of both the central and lateral target locations (see Figure 1). The inset in Figure 1 shows the per-stride average head yaw trajectoiy for the central target portion of the trial. The desired 25,50,75, and 100% points for the possible target relocation times are depicted on the plot. ........................................................................................... Insert Figure 1 here ........................................................................................... In all subjects, the medio-lateral position of a marker located near the seventh cervical vertebra maintained an anti-phase relationship with the head yaw signal. Across all of the data trials the peak-to-peak amplitude of the lateral trunk translation was 4.9 -+ 1.0 cm (mean -+ 1 SD) (range: 3.1 to 5.6 cm) and the peak-to-peak amplitude of the head yaw rotation was 3.2 1.2 degrees 2 (range: 2.1 to 5.1 degrees). Four of the six subjects also had horizontal eye movements that clearly showed a similar oscillatoiy pattern that maintained an anti-phase relationship with the head yaw rotation (see Figure l),i ndicating that when the head rotated to the left, the eyes rotated to the light. This type of coordination could be consistent with a natural head oi-ientation projection that crosses the standing line of sight between the subject and the visual target In this scenario, the counter- 15. 7 rotation of the eyes is necessary because the head rotation over-compensates for the lateral '' translation of the hxnk 16. Reaction Time Variables With subject means ranging from 209 to 288 ms (group mean: 252 ms), the horizontal eye movement onset latencies varied significantly between subjects (p < 0.001). This was also true for the latency of the onset for the head yaw movement (range: 234 to 290 ms; group mean: 262 ms). The group means suggest that the onset of the eyes preceded the head movement onset, but this was only clearly evident for half of the subjects (see Figure 2). A presentation of the p-values obtained for each of the dependent variables across all experimental manipulations is provided in Table 1. It is evident from these data that neither of the reaction time variables were affected by the experimental manipulations ........................................................................................... Insert Figure 2 here ........................................................................................... ........................................................................................... Insert Table 1 here Head Saccade Dynamics The saccade amplitude and peak velocity of the head yaw signal were measured to provide an assessment of head saccade dynamics. Across all subjects the mean peak velocity values ranged from 64 - 172"h and the amplitude of the head saccade ranged from 3-25". Each of these measures illustrate that the head dynamics of the saccade response differed between subjects (p < 0.001 for each). A statistically significant interaction between the direction of the required gaze re-fixation and the point in the gait cycle at which the target changed locations was also observed in each @=0.015a nd p = 0.021 for peak velocity and saccade amplitude, respectively). The presence of this interaction, clearly visible in the group means presented in Figure 3, indicates that stride cycle parameters can influence the head movements made to acquire visual targets while walking. 8 ........................................................................................... Insert Figure 3 here ........................................................................................... A reason for this interaction becomes apparent when inspecting a subject’s head yaw velocity signals from the 25 and 75% conditions (see Figure 4). A comparison of the peak velocity amplitudes between the two conditions shows that movements made to fixate a target in the left periphery achieve higher magnitudes in the 75% condition than those in the 25% condition. The opposite is true for rightward saccades. The slope of the head velocity signal just piior to the onset of the movement shows the direction of the head’s rotational acceleration at the time that the movement is initiated. An upward slope indicates that the head is accelerating to the left. When the visual target appears in the periphery to the side that the head is naturally accelerating, the accompanying head movement saccade achieves a higher peak velocity. When required to stop and accelerate in the opposite direction the magnitude of the peak velocity of the saccade is lower. These results indicate that the natural motions of the head that occur while walking affect the dynamics of the head saccade. ........................................................................................... Insert Figure 4 here ........................................................................................... Discussion The results presented here suggest that the head movement responses to visual target position changes can be influenced by the phase of the gait cycle during which the target relaxtion takes place. Differences were present in the peak head velocity and head saccade amplitude signals, but the movement onset latencies of the eye and head movements appeared to be unaffected by the experimental manipulations. The fact that the reaction time variables (i.e. the eye and head movement onset latencies) were unaffected by the experimental conditions of this walking paradigm is in contrast with ’’* previously repoi-ted results of reaction time tasks peiformed while walking . The previous 17, 9