Use of monocular and binocular visual cues for postural control in children

Abstract

The purpose of the present study was to examine how children's postural control is affected by different visual cues. Sixteen children, ages 8 and 12 years, and eight adults participated in the study. Each was asked to stand upright inside a moving room, which oscillated at 0.2 Hz, while facing the frontal wall at two distances: 75 and 150 cm, under monocular and binocular vision conditions. Vision manipulation induced corresponding body sway in all participants, but vision effect was the smallest in the monocular vision condition, at the greater distance (150 cm) from the front wall. More importantly, however, the influence of visual manipulation on body sway was age-dependent, with the younger children showing less visually induced body sway than the older children and adults. This aging effect was more dramatic in the monocular vision condition. These results suggest that development of the visual system is not fully completed until the age of 12 years and that eye movement and binocular vision might play an important role in how visual cues are coupled to body sway.

Introduction

Postural control relies upon multiple sources of sensory information, which originates mainly in the visual, vestibular, and somatosensory systems. Postural orientation and equilibrium are achieved by the integration of sensory inputs from all sensory systems (Horak & Macpherson, 1996). Yet, many studies about visual influences on postural stability have revealed that the simple act of closing the eyes causes a decrease in upright stability. When the eyes are open, body oscillation causes the image on the retina to slip, which automatically initiates the appropriate muscle activation to minimize body sway (Paulus, Straube, Krafcyk, & Brandt, 1989). Recently, it has been suggested that afferent ocular motor signals also play a major role in the control of body orientation (Kapoula & Lê, 2006; Strupp et al., 2003). However, the use of visual cues to control posture is not fully understood. 

Using a moving room, Moraes, Lopes, and Barela (2009) confirmed that binocular vision provides more information as compared to monocular vision and that this effect is smaller when visual cues are modified via changes of distance in the moving room. These results were explained through retinal slip and its accompanying motor compensation, as described by Paulus et al. (1989). As the distance from the moving scene increases, the rate of expansion/contraction projected on the retina diminishes, since retinal slip decreases, and consequently, body sway is hardly detectable. Additionally, with monocular vision, both eyes still converge, although more weakly (Busettini, Masson, & Miles, 1997), and consequently, smaller eye movement is produced and less afferent ocular motor signals are available. Fewer eye movements in a moving scene results in less influence by the visual manipulation and less sway induced by the moving scene. 

Godoi and Barela (2008) found that children of up to 10 years of age, who were exposed to a moving scene as they stood upright, were less affected by the increase of the distance between them and the moving scene than were their older peers and adults. These results suggest that changes in retinal slip are less detectable by young children and that adult-like behavior is observed only in 10-year-old and older children (Godoi & Barela, 2008). The understanding of children's sensitivity to a moving room (global motion) is controversial. For example, Ellemberg et al. (2004) reported that adult-like coherence motion thresholds occur by 6 years of age, whereas Gunn et al. (2002) reported that this occurs in 10- to 11-year-olds. However, as noted by Lewis and Maurer (2005), Gunn et al. used stimuli with a speed lower than that used by Ellemberg et al. Since the moving room speed was quite low, it is possible that, in this case, the young children were not able to detect its movement at a far distance, because retinal slip was diminished. Finally, Greffou, Bertone, Hanssens, and Faubert (2008) identified that, in fully immersive virtual reality, adult-like postural response as related to magnitude and variability was observed only in 16-year-old individuals. 

Developmental changes have been observed in other aspects of the visual system. While vision acuity reaches an adult-like level by the age of 6 years (Ellemberg, Lewis, Liu, & Maurer, 1999), stability of the eyes immediately following saccades is still poor in children around this age, with dramatic improvement occurring around the age of 7 or 8 years for far distances. Adult-like values for near vision occur at 10 to 12 years (Yang & Kapoula, 2003). It has also been shown that the maturation of visual regions in the central nervous system that are involved with the processing of different vision attributes may occur well into childhood and even adolescence (Klaver et al., 2008; Lichtensteiger, Loenneker, Bucher, Martin, & Klaver, 2008). A seemingly interesting result of these studies is that the process of functional segregation into ventral and dorsal streams does not mature by the age of 6 years for biological motion or from structure-from-motion stimuli. While adults exhibited neural specialization within the dorsal stream areas, the children activated both ventral and dorsal areas, suggesting less-apparent neural specialization. Considering that optic flow stimuli are processed in the dorsal stream (Ohlendorf, Sprenger, Speck, Haller, & Kimmig, 2008; Smith, Wall, Williams, & Singh, 2006), it is possible to assume that optic flow perception can be affected in preschool age children, since dorsal stream maturation seems to take longer than does ventral stream maturation (Gunn et al., 2002). 

These changes in vision development lead to questions about how children's postural control is affected by different visual cues: that is, how visual cues and body sway are coupled. It has been suggested that adults, due to their mature visual systems, are less affected by changes in the quality of visual cues, as observed in body sway induction due to a moving scene (Godoi & Barela, 2008). The purpose of the present study, therefore, was to examine the coupling between visual information and body sway in children, using binocular and monocular vision, at two distances from the frontal wall of a moving room. We hypothesized that children would not compensate, as do adults, to changes in distance from the moving scene and that the change from binocular to monocular vision would more greatly affect the performance of the postural control system of children than of adults. 

Methods

Subjects

Sixteen typically developing children and eight healthy adults (21.75 ± 2.49 years) participated in this study. The groups of children were comprised of 8- (n = 8, 8.46 ± 0.34 years) and 12-year-olds (n = 8, 12.44 ± 0.40 years). Children were recruited from the community through contacts with friends and families. Adult participants were graduate and undergraduate students. Participants did not have any health problems that could affect their performance. Adults and children's parents/guardians signed the informed consent form approved by the Institutional Ethics Committee at São Paulo State University before engaging in the study. 

Procedures

Participants were asked to stand as still as possible inside of a moving room, while facing the front wall. This room consisted of three walls (2.1 m long, 2.1 m wide, and 2.1 m tall) and a ceiling and was mounted on wheels. The room oscillated in the anterior–posterior (AP) direction by a servomotor mechanism. The ceiling was coated with white paint, and the walls were painted with a pattern of alternating white (42 cm wide) and black (22 cm wide) stripes. The servomotor mechanism consisted of a controller (Model APEX 6151, Compumotor, CA, USA), a controlled stepper motor (Compumotor, CA, USA), a software program that controlled the motor inputs (Motion Architect for Windows, Compumotor, CA, USA), and an electrical cylinder (Compumotor, CA, USA) that connected the servomotor mechanism to the moving room. Although the whole moving scene could be moved, the floor was kept stationary. A fluorescent lamp (20 W) was placed on the room's ceiling to keep the room illuminated by constant luminescence. 

Participants stood in the middle of the moving room with respect to the sidewalls and at two distances from the front wall (75 and 150 cm), under two visual conditions (monocular and binocular). For the binocular and monocular conditions, the subtended visual angle of the front wall was equal to 109° and 70°, at distances of 75 and 150 cm, respectively. For the sidewalls in the binocular condition, the subtended visual angles were calculated based on a field of view of 200° (Mon-Williams, Wann, & Pascal, 1999). The values of the sidewall subtended visual angle were 45.5° and 65° (91° and 130° when adding both sides), at 75 and 150 cm, respectively. For the monocular condition, the subtended visual angles of the sidewalls were calculated based on a field of view for one eye, which was 160°, as suggested by Good, Fogt, Daum, and Mitchell (2005). The subtended visual angle was equal to 51° (45.5°, plus 5.5° from the occluded side) and 90° (65°, plus 25° from the occluded side) at 75 and 150 cm, respectively. Details about the calculation of these subtended visual angles can be obtained in Moraes et al. (2009). 

Participants performed the monocular vision condition with their dominant eye, with the non-dominant eye covered with an adapted sleep mask. The dominant eye was determined by first, with binocular vision, placing a finger in front of a target 5 m away and then asking which eye during monocular vision provided the closest positioning of the finger to the target. The combination of distance and visual conditions resulted in four experimental conditions and a total of 12 trials, each lasting 60 s, and distributed in three blocks. In the first block of trials, the participants performed one trial for each experimental condition, with the room kept stationary. In the subsequent two blocks of trials, the participants performed the four experimental conditions, combined with moving room movement. In all blocks, the order of the four experimental conditions was randomized. 

While inside the moving room, participants were requested to stand barefoot with feet placed comfortably apart and to keep the upper limbs relaxed alongside the body. During each trial, participants were asked to fixate their gaze on a target located at eye level on the front wall. Participants had no experience of, or previous information about, the movement of the room, since this knowledge might influence postural responses to visual manipulation (Freitas Júnior & Barela, 2004). The frequency of the moving room was set at 0.2 Hz, which corresponds to the natural frequency oscillation of the human body in the standing upright position (Soames & Atha, 1982). In order to achieve this frequency, the room was moved at a peak velocity of 0.6 cm/s and amplitude of 1.2 cm. 

An infrared emitting diode (IRED) was placed centrally on the back of each participant at the scapula level (∼8th thoracic vertebra). Another IRED was placed on the front wall of the moving room. One OPTOTRAK camera (Northern Digital, Waterloo, Canada) positioned behind the participant tracked the IREDs at a sampling rate of 100 Hz. Kinematic data were filtered using a fourth-order, zero lag, low-pass digital Butterworth filter, with a cutoff frequency of 4 Hz. 

Data analysis

Since the room moved in the AP direction, the variables used in this study were computed only for this direction. In addition, due to our interest in the coupling between body sway and moving room movement, coherence, gain, phase, position, and velocity variability were calculated for each trial in which the room was oscillated and then averaged across groups, visual conditions, and distance from the scene. A frequency-response function (FRF) was computed from body sway and visual stimulus for each participant and trial. FRF was determined by dividing the Fourier transforms of body oscillation by the Fourier transforms of the visual stimulus, generating a complex-valued function. 

Coherence measured the strength of the relationship between body sway and movement of the moving room: that is, how strongly body sway was coupled to the visual stimulus. Coherence values close to 1 indicated that the signals demonstrated a strong dependency between them. Conversely, values close to zero indicated that the signals demonstrated a weak or no dependency between them. Gain was computed as the absolute value of the transfer function and indicated the coupling strength between visual stimulus and body sway. A gain value of 1 indicated that the spectrum amplitude of body sway was equal to the spectrum amplitude of the room's movement. Phase was computed as the argument of the transfer function, converted in degrees, and indicated the temporal relationship between visual stimulus and body sway. Positive values for phase indicated that body sway led the room's movement; negative values indicated that body sway lagged behind the room's movement. 

Position and velocity variability of body sway were computed as the standard deviation of body sway trajectory and velocity (Jeka, Oie, & Kiemel, 2000) after the component of body sway due to stimulus frequency was removed (residual trajectories). Position and velocity variability values indicated body sway amplitude and velocity (sway variability), respectively, at frequencies other than the 0.2-Hz frequency, and with higher values indicating higher variability. 

In addition, mean sway amplitude was computed for the trials in which the room was kept stationary by calculating the standard deviation of the body sway time series, after the average of body sway position was subtracted from the data points within each trial and then averaged across groups. Mean sway amplitude indicated the overall body sway throughout each trial, with lower values indicating better maintenance of an upright stance. 

Statistical analysis

Two ANOVAs (group × distance × vision), with repeated measure in the last two factors, were carried out to evaluate the effects of group (8- and 12-year-olds and adults), distance (75 and 150 cm), and vision (monocular and binocular) on mean sway amplitude and coherence. In addition, two MANOVAs (group × distance × vision), with repeated measure in the last two factors, were carried out to evaluate the effects of group, distance, and vision on (1) gain and phase and (2) position and velocity variability. Appropriate follow-up univariate and post-hoc analyses with Bonferroni adjustments were performed, when necessary. The significance level was 0.05. All analyses were performed using the SPSS package (SPSS version 10.0). 

Results

When the room remained stationary, ANOVA revealed no main effect for group and vision; however, it revealed that mean sway amplitude was larger when the individual stood at the 75-cm distance from the wall than at the 150-cm distance, F(1, 21) = 4.99, p < 0.05. Participants' body oscillation was induced by visual manipulation, with sway observed at the same frequency as the driving signal, across all trials. The results of the specific variables are reported below. 

Coherence

ANOVA revealed main effect for group, F(2, 21) = 23.04, p < 0.001, vision, F(1, 21) = 76.52, p < 0.001, and distance, F(1, 21) = 14.91, p < 0.005, as well as for group by vision interaction, F(2, 21) = 9.17, p < 0.005. For the group main effect, Bonferroni post-hoc analysis identified that all age groups were different from each other, in such a way that coherence increased as the individuals aged (8-year-old children = 0.59∣12-year-old children = 0.72∣Adults = 0.84). Coherence was greater for the binocular condition (0.82) than for the monocular condition (0.62). Coherence also was greater for the 75-cm distance (0.75) than for the 150-cm distance (0.70). For the interaction effect, multiple t-tests, with Bonferroni adjustment (6 comparisons: p ≤ 0.0083), were carried out as post-hoc analyses. For the binocular condition, coherence was smaller for the 8-year-olds as compared to the adults (p = 0.0061). For the monocular condition, coherence was smaller for both the 8-year-olds and the 12-year-olds than for the adults (p = 0.00001 and p = 0.0007, respectively; Figure 1). 

Figure 1

 

Mean coherence values between body sway and moving room position for each age group (8- and 12-year-olds, and young adults) across monocular and binocular vision conditions and 75 and 150 cm from the frontal wall of the moving room.

Gain and phase

Figure 2 depicts gain and phase values for all three groups and for vision and distance conditions. MANOVA revealed main effect for vision, Wilks' Lambda = 0.143, F(2, 20) = 59.84, p < 0.001, distance, Wilks' Lambda = 0.274, F(2, 20) = 26.52, p < 0.001, and vision by group interaction, Wilks' Lambda = 0.603, F(4, 40) = 2.87, p < 0.001. The follow-up univariate analyses revealed that the main effect of vision, F(1, 21) = 125.26, p < 0.001), distance, F(1, 21) = 55.04, p < 0.001), and group by vision interaction, F(2, 21) = 5.91, p < 0.01, occurred only for gain. In general, gain was higher for the 75-cm distance (0.90) than for the 150-cm distance (0.66) and lower for the monocular condition (0.62) in comparison with the binocular condition (0.94; Figure 2). The multiple t-tests with Bonferroni adjustment, used as post-hoc analyses, failed to identify any difference among age groups in both monocular and binocular conditions. 

Figure 2

 

(a) Mean gain and (b) phase values between body sway and moving room position for each age group (8- and 12-year-olds, and young adults) across monocular and binocular vision conditions and 75 and 150 cm from the frontal wall of the moving room.

Position and velocity variability

Figure 3 depicts position and velocity variability for all three groups and for vision and distance conditions. MANOVA revealed main effect for group, Wilks' Lambda = 0.501, F(4, 40) = 4.13, p < 0.01, vision, Wilks' Lambda = 0.605, F(2, 20) = 6.52, p < 0.01, and distance, Wilks' Lambda = 0.526, F(2, 20) = 9.01, p < 0.005. Univariate analyses revealed significant main effects for group, F(2, 21) = 3.63, p < 0.05; F(2, 21) = 9.51, p < 0.005, vision, F(1, 21) = 9.22, p < 0.01; F(1, 21) = 9.71, p < 0.01, and distance, F(1, 21) = 5.14, p < 0.05; F(1, 21) = 18.906, p < 0.001, for both position and velocity variability. In general, position and velocity were higher for the 75-cm distance (0.64 and 0.62) than for the 150-cm distance (0.58 and 0.57) and higher for the monocular condition (0.64 and 0.61) than for the binocular condition (0.58 and 0.58; Figure 3). Post-hoc analyses revealed that position and velocity variability values were also age-dependent, with the 8-year-old children showing higher values than the adults (p = 0.044 and p = 0.001). No differences were found between the 8- and 12-year-old children and the 12-year-old children and adults. 

Figure 3

 

(a) Mean position and (b) velocity variability values between body sway and moving room position for each age group (8- and 12-year-olds, and young adults) across monocular and binocular vision conditions and 75 and 150 cm from the frontal wall of the moving room.

Discussion

The goal of this study was to examine the coupling between visual information and body sway in children, using monocular and binocular vision, at two distances from the frontal wall of a moving room. Our results revealed that children were able to couple to the visual information, since they displayed corresponding body oscillations to the visual cues. However, while this coupling was based on the same underlying parameters as for the adults, it was not as well defined and highly tuned as was observed in the adults. Because this coupling is not completely developed in children, changes in the quality of the visual cues—especially those that changed from binocular to monocular vision—most affected the children, particularly the younger ones. 

Corresponding body oscillation due to visual manipulation has been observed in sitting (Barela, Godoi, Freitas Junior, & Polastri, 2000; Bertenthal, Boker, & Xu, 2000) and standing infants (Delorme, Frigon, & Lagacé, 1989) and in children (Godoi & Barela, 2008; Rinaldi, Polastri, & Barela, 2009; Schmuckler, 1997), in real moving rooms as well in virtual reality environments (Greffou et al., 2008). In addition, based on the gain and phase values, which were similar between the adults' and children's responses to the visual cues, it appears that the processes underlying such behaviors are similar between infants, children, and adults. Therefore, the coupling between visual information and postural responses seems to constitute a basic mechanism, which is already in place after just the first few days of life (Jouen, Lepecq, Gapenne, & Bertenthal, 2000). However, as observed in the current results as well as in others (Godoi & Barela, 2008; Greffou et al., 2008; Rinaldi et al., 2009; Schmuckler, 1997), it seems that this mechanism is refined during the first years of life. 

The main question of the present study was whether or not the availability of different visual array cues—the result of manipulation of monocular and binocular conditions and at different distances from the frontal wall of a moving room—would change the postural responses in children. We demonstrated that monocular vision, combined with the distance that was furthest from the frontal wall, led to the smallest response from the children, whereas, in the adults, the binocular condition, combined with the closer distance, led to a greater response (Moraes et al., 2009). Additionally, in an earlier study, the manipulation of distance produced a decreased response to the visual moving room only in children younger than 10 years old (Godoi & Barela, 2008). 

The current results indicate that the children's postural responses were more affected by the manipulation of visual cues than were the adults' responses, implying that coupling between visual information and body sway is not fully developed until the age of 12 years. Coherence values observed for the children were lower than those of the adults, corroborating previous observations in a virtual reality environment (Greffou et al., 2008). However, our data indicate that the 12-year-old children seemed to be in a transitional period, in which similar differences for the adults were observed only in the monocular vision condition. Conversely, the 8-year-old children showed lower coherence values as well as higher variability (position and velocity) values, in both monocular and binocular conditions. Interestingly, for coherence, the difference between the 8-year-old children and the adults was much greater in the monocular vision condition. A trend was observed in coherence values in the monocular vision condition, which increased from the 8-year-old to the 12-year-old children and then to the adults (Figure 1). Such a trend was not present in the binocular vision condition, in which the 12-year-old children's responses were similar to those of the adults. 

Additionally, the increase in distance from the frontal wall influenced the strength of this relationship, but this manipulation was not as affective as the vision manipulation. Monocular vision clearly disrupted the more coherent body sway responses, more so than did increasing the distance to the frontal wall. Accordingly, the monocular vision was less informational as was binocular vision, and the use of monocular vision was age-dependent. Eye convergence can be altered through the use of one or both eyes (Busettini et al., 1997), and it can alter the quality of afferent ocular motor signals, reducing coupling between visual information and body sway. This was observed in the younger children in the monocular vision condition. Moreover, children's ability to stabilize the eyes after saccades is still developing up to the end of the first decade of life (Yang & Kapoula, 2003). Eye vergence in children also may be less stable, which, when combined with monocular vision, would further affect it and, therefore, disrupt body response even more in the monocular condition. 

Although the youngest age group tested in the present study was 8-year-olds, the results support findings for the immature state of the dorsal stream in 6-year-old children (Gunn et al., 2002; Klaver et al., 2008; Lichtensteiger et al., 2008). Since specific regions in the parietal cortex are sensitive to optic flow stimuli (Ohlendorf et al., 2008; Smith et al., 2006), it could be expected that children would exhibit a less strong coupling between visual manipulation and body sway. Consequently, less activation of optic flow-sensitive regions in the dorsal stream could lead to a reduced perception of environmental motion, which in turn would dampen body sway. 

Relative to monocular vision, it is known that unilateral flow field stimulation exclusively activates the middle temporal (MT) area in the dorsal stream, whereas the medial superior temporal (MST) area is sensitive to both unilateral and bilateral flow field stimulation (Ohlendorf et al., 2008). This specific segregation into the MT and MST areas for unilateral and bilateral stimulation, which are similar to monocular and binocular vision, respectively, may also still be immature in 8-year-olds. Therefore, for the youngest age group tested in the present study, monocular perception of optic flow may have been more compromised than for the others, which may have additionally disrupted the coupling between the moving room stimulus and body sway. 

In summary, children coupled to visual manipulations of the surroundings with similar properties of those observed in adults but with less strength and stability. As a consequence, any manipulation of the quality of visual cues, especially the change from binocular to monocular visual condition, most disrupted the younger children. These results indicate that coupling between visual information and body sway is not fully developed up to the age of 12 years and might reflect the immature state of the dorsal stream. 

Acknowledgments

J. A. Barela is grateful to São Paulo Research Foundation (FAPESP, Brazil; Financial Support, Grant No. 2009/16051-8). A. G. Lopes is grateful to CNPq (Post Doctoral Support, Grant No. 57105/2010-5). M. Razuk is grateful to CNPq (Master's Assistantship, Grant No. 145789/2010-1). R. Moraes is grateful to São Paulo Research Foundation (FAPESP, Brazil; Financial Support, Grant No. 2009/14102-4). 

Commercial relationships: none. 

Corresponding author: José A. Barela. 

Email: [e-mail]. 

Address: Instituto de Ciências da Atividade Física e Esporte (ICAFE), Cruzeiro do Sul University, Rua Galvão Bueno, 868, São Paulo, SP 01506-000, Brazil.