- Email: [email protected]
Development of adaptive sensorimotor control in infant sitting posture
Gait & Posture 45 (2016) 157–163
Contents lists available at ScienceDirect
Gait & Posture journal homepage: www.elsevier.com/locate/gaitpost
Development of adaptive sensorimotor control in infant sitting posture Li-Chiou Chen a,b,*, John Jeka c,d,e, Jane E. Clark c,d a
School and Graduate Institute of Physical Therapy, National Taiwan University, Taipei, Taiwan Physical Therapy Center, National Taiwan University Hospital, Taipei, Taiwan c Department of Kinesiology, University of Maryland, College Park, MD, USA d Graduate Program in Neuroscience and Cognitive Science, University of Maryland, College Park, MD, USA e Department of Kinesiology, Temple University, Philadelphia, PA, USA b
A R T I C L E I N F O
A B S T R A C T
Article history: Received 28 February 2015 Received in revised form 4 November 2015 Accepted 12 January 2016
A reliable and adaptive relationship between action and perception is necessary for postural control. Our understanding of how this adaptive sensorimotor control develops during infancy is very limited. This study examines the dynamic visual–postural relationship during early development. Twenty healthy infants were divided into 4 developmental groups (each n = 5): sitting onset, standing alone, walking onset, and 1-year post-walking. During the experiment, the infant sat independently in a virtual moving-room in which anterior-posterior oscillations of visual motion were presented using a sum-of-sines technique with five input frequencies (from 0.12 to 1.24 Hz). Infants were tested in five conditions that varied in the amplitude of visual motion (from 0 to 8.64 cm). Gain and phase responses of infants’ postural sway were analyzed. Our results showed that infants, from a few months post-sitting to 1 year post-walking, were able to control their sitting posture in response to various frequency and amplitude properties of the visual motion. Infants showed an adult-like inverted-U pattern for the frequency response to visual inputs with the highest gain at 0.52 and 0.76 Hz. As the visual motion amplitude increased, the gain response decreased. For the phase response, an adult-like frequency-dependent pattern was observed in all amplitude conditions for the experienced walkers. Newly sitting infants, however, showed variable postural behavior and did not systemically respond to the visual stimulus. Our results suggest that visual–postural entrainment and sensory re-weighting are fundamental processes that are present after a few months post sitting. Sensorimotor refinement during early postural development may result from the interactions of improved self-motion control and enhanced perceptual abilities. ß 2016 Elsevier B.V. All rights reserved.
Keywords: Posture Vision Sensorimotor Infant Sitting
1. Introduction Postural control is an important motor skill acquired during early development. To control the multi-segmented body in various conditions, a reliable and adaptive relationship between action and perception is necessary. Evidence of adaptive sensorimotor control in posture has been shown in adults [1,2] and in children [3–5]. Little is known about how this type of control develops during infancy. Young adults entrain their standing posture to the frequency properties of vision and somatosensory information [1,6,7]. The
* Corresponding author at: School and Graduate Institute of Physical Therapy, National Taiwan University, No. 17, Xuzhou Rd., Taipei 100, Taiwan. Tel.: +886 2 33668135; fax: +886 2 33668161. E-mail address: [email protected] (L.-C. Chen). http://dx.doi.org/10.1016/j.gaitpost.2016.01.020 0966-6362/ß 2016 Elsevier B.V. All rights reserved.
frequency response of this sensory-postural relationship demonstrates an inverted-U pattern with the greatest in-phase entrainment near 0.2 Hz and weaker and out-of-phase entrainment as the stimulus frequency decreases or increases [6,7]. Furthermore, adults’ postural sway is proportional to the stimulus amplitude within a range. When the amplitude exceeds a certain value, postural response decreases [2,8]. This change in the postural response to a change in the sensory stimuli indicates sensory re-weighting, a critical component of the adaptive sensorimotor control [1,2,9]. When a source of sensory information is unreliable, the postural system needs to attenuate this source of information and increase reliance on another modality. This sensory re-weighting process has been shown in children as young as 4 years of age [3] but has not been investigated in very young children or infants. Newborn infants show directionally appropriate and velocityscaled head response to visual flow information [10], suggesting
158
L.-C. Chen et al. / Gait & Posture 45 (2016) 157–163
that the visual–postural relationship may be a fundamental component of the sensorimotor system. A few studies have investigated infants’ postural responses to changes in the frequency of visual stimuli [11–13]. However, the results were conflicting. While some studies showed that infants’ postural entrainment increased as the frequency of the visual motion increased [11,12], one study reported a linear decline in sway coherence as a function of stimulus frequency [13]. Furthermore, the developmental changes in infants’ visual–postural coupling were inconsistent between the studies [11,12]. These conflicting results may be due to differences in the postural measures as well as the range of sensory properties employed in the studies. Since children were shown to respond to a wider frequency range of visual stimuli compared to adults [5], a sufficient range of sensory properties is required to better characterize the sensorimotor relationships in infants’ postural control. In the present study, we sought to systematically examine the dynamic visual–postural relationship in infants as they develop postural control. Specifically, we examined infants’ ability to adapt their sitting posture to different properties of the visual signal, i.e., frequency and amplitude. Incorporating a wide range of input frequency and amplitude, we addressed the following research questions: (1) does the adult patter (inverted-U) of visual–postural coupling exist in early infancy? (2) is sensory re-weighting a fundamental process of postural control present in young infants? Using a cross-sectional design, we also examined how the adaptive visual–postural relationship may differ by infants’ advances and experience in postural development from sitting to standing and finally to walking.
2. Method 2.1. Participants Twenty healthy infants (13 boys) composed four groups (each n = 5): (1) sitting onset (SO; aged 6.7 1.1 months; 6.0 5.1 days after sitting onset), (2) standing alone (ST; aged 10.6 1.2 months; 28.8 11.9 days before walking onset), (3) walking onset (WO; aged 11.7 1.4 months; 8.8 8.1 days after walking onset), and (4) 1-year post-walking (W12; aged 23.5 1.2 months; 11.7 1.0 months after walking onset). Sitting onset was when the infant could sit without support for 10 s. Standing alone was when the infant first able to stand without support for 5 s. Walking onset was when the infant started to walk independently for three continuous steps. Although no formal developmental assessment was done, the first author, a pediatric physical therapist, screened all infants to assure ageappropriate development. All experimental procedures were approved by the Institutional Review Board of the University. Parents gave written informed consent before participation.
Fig. 1. An infant sits independently on a chair in a 3-wall room. The smiley face represents the location of the cartoon video.
Infants’ postural sway was measured using an active infrared position tracking system (Optotrak, Northern Digital Inc.). Three small infrared LEDs were affixed to the infant’s occipital prominence, upper and middle trunk. A bank of three cameras was positioned parallel to the front screen and 2 m behind the infant. In addition, the testing session was videotaped for later coding of the infant’s sitting behavior. 2.3. Experimental design and procedures The visual stimulus was created by anterior–posterior (AP) oscillations of the animated display using a sum-of-sines technique [14] that allowed us to examine infants’ visual–postural coupling over a large range of frequency and amplitude variations without increasing the testing burden. The sum-of-sines consisted of a summation of 5 sinusoids at frequencies 0.12, 0.28, 0.52, 0.76, and 1.24 Hz with baseline amplitudes of 0.417, 0.179, 0.096, 0.065, and 0.040 cm, respectively. Five amplitude conditions were tested: A0: The visual display was stationary. A1: The visual display oscillated with the baseline amplitudes as described above. Peak-to-peak amplitude was 1.44 cm. A2: The amplitudes were twice of those in A1 condition. A4: The amplitudes were four times of those in A1 condition. A12: The amplitudes were twelve times of those in A1 condition. During the experiment, one experimenter and the parent stayed near the infant but not within his/her sight to provide help when needed. Data were collected in 3 randomized blocks, each with one 60-s trial for each amplitude condition.
2.2. Apparatus and measures 2.4. Data reduction and analysis Fig. 1 illustrates the experimental set-up in which the infant sat on a customized chair that was fixed on a 45 cm high pedestal and placed 100 cm from the front wall. The chair had a small back support and a safety belt loosely across the infant’s hip. The visual stimulus was created in a Fakespace Systems CAVE Automatic Virtual EnvironmentTM that is a rear-projected 3-screen display (each 2.5 m 3.0 m) with 1280 1024 pixels spatial resolution and 60 Hz framing rate. The visual display was specified by white triangles (0.28 0.38 0.28) randomly projected on a black background, excluding the foveal region (15 cm diameter circle) on the front screen. To attract the infant’s attention to the front screen, a cartoon video was projected onto the foveal region with auditory outputs behind the front screen.
Videos of all sitting trials were coded for usable time segments. In this study, the shortest period required to characterize all sum-of-sines components is 8.33 s (for the lowest frequency 0.12 Hz). For comparison with previous studies [15,16] and for coding convenience, we decided the minimal duration for sitting time segments as 10 s. The coding criterion was that the infant maintained quiet posture and continuous visual engagement to the front screen for at least 10 s. Only those time segments were used for subsequent data analysis. Mean segment time (MST) was computed across all segments of each infant in each amplitude condition to represent the infant’s engagement in the sitting task.
L.-C. Chen et al. / Gait & Posture 45 (2016) 157–163
Infants’ postural sway was analyzed only in the AP direction as it was the direction of visual stimulus. Head and trunk displacements were mean-detrended and low-pass filtered (recursive 2ndorder Butterworth filter, fcut-off = 5 Hz). For infants’ postural sway at the driving frequencies, the transfer (frequency-response) function at each frequency was computed dividing the Fourier transform of the postural response by that of the stimulus. Gain and phase were derived from the mean transfer function [7]. Gain represents the strength of the postural response to the stimulus. Larger gain values suggest more responsive to the stimulus. Phase is normalized time delay between the postural sway and the stimulus motion. A positive phase value indicates that the body movement leads the visual stimulus. Phase values that are not tightly clustered suggest inconsistent response to the stimulus. For infants’ postural responses at the non-driving frequencies, mean position variability (Pos_var) and velocity variability (Vel_var) were calculated for the residual postural sway, with the components of all 5 driving frequencies removed. To characterize infants’ quiet sitting posture, postural sway in the A0 condition was examined using distance-related and raterelated measures. Distance-related measures, including Pos_var, Vel_var, and mean velocity (Vel_m), were calculated to represent postural steadiness. Rate-related measures, including velocity and frequency variables, were used to indicate changes of postural control strategies [17,18]. For sway frequency, power spectrum density of postural sway time series was computed using multitaper method with 8 tapers and median frequency (F_m) was calculated. Mean for each postural measure was calculated, weighted by the segment duration. 2.5. Statistical analysis To examine infants’ postural responses to the moving visual stimulus, transfer functions were compared in the complex plane using a linear model of repeated measures MANOVA. The dependent variables were the real and imaginary parts and the independent variables were frequency, amplitude, and group. This analysis assessed the transfer function distributions among the independent variables and therefore took into account both gain and phase. Transfer functions with different gain responses but large phase variability may not be seen as different in this analysis. To further examine infants’ spatial and temporal postural responses to the visual stimulus, gain and phase were separately examined using mixed model repeated measures ANOVAs. To assess the effects of infant group and stimulus amplitude on the residual postural sway and to examine whether infants’ unperturbed sitting posture differed among the four groups, mixed model repeated measures ANOVAs were conducted. All statistical analyses were performed using the Statistical Analysis Software (SAS) program (Release 9.1, SAS Institute Inc.). The a level was set at 0.05 and post hoc comparisons with Tukey adjustment were performed when applicable. 3. Results All infants completed the experiment except that one sitting onset infant was unable to remain sitting in the A12 condition. MST results showed a significant group effect (p < 0.05). One-year postwalking infants engaged in the sitting task longer than all other infants (MST 36.5 8.9 and 25.8 15.6 s, respectively). No significant difference was found among amplitude conditions or the three younger groups. Infants’ postural responses at head, upper and middle trunk showed similar patterns among the groups, amplitude conditions, and frequencies. As would be expected for segments more distal to the support surface, the postural response was larger for the
159
head than for the trunk. Therefore, our current analysis focused on results from head data. 3.1. Postural responses to the visual stimulus: frequency and amplitude responses The transfer functions for infants’ postural responses to the visual stimulus at the driving frequencies did not cluster on the complex plane (see exemplars in Fig. 2). Repeated measures MANOVA revealed a significant frequency effect (Wilks’ Lambda = 0.79, p < 0.05) and a marginal group frequency effect (Wilks’ Lambda = 0.89, p = 0.064). Significant frequency effects were found in all infant groups except for sitting onset infants. Gain and phase of infants’ postural responses to the visual stimulus were further separately analyzed. For gain, significant effects of frequency, amplitude, group, and group amplitude were found (all p < 0.001). All infants showed significantly higher gain responses at 0.52 and 0.76 Hz than at 0.12 and 1.24 Hz. As the stimulus amplitude increased, gain response decreased (Fig. 3). A significant group effect existed only in the A1 condition in which sitting onset infants showed higher gains than all other infants. For phase, significant frequency, frequency group, frequency amplitude, and frequency amplitude group effects were found (all p < 0.05). All infants demonstrated positive phase (posture leading stimulus) at lower frequencies and gradually became negative phase (lagging behind) at higher frequencies. One-year post-walking infants showed significant phase differences between 0.52 and 1.24 Hz in all amplitude conditions while sitting onset infants demonstrated no frequency response in any amplitude condition. For infants’ residual postural sway at the non-driving frequencies, a significant group effect was revealed for both Pos_var and Vel_var (p < 0.001). Sitting onset infants demonstrated significantly higher Pos_var than standing alone and 1-year post-walking infants, and walking onset infants higher than 1-year post-walking infants. Vel_var was highest in sitting onset infant and lowest in 1year post-walking infants (Fig. 4). 3.2. Unperturbed sitting posture For infants’ unperturbed sitting posture, all sway measures in the A0 condition showed significant group effects (all p < 0.05; Fig. 5). Pos_var was larger in sitting onset infants than in standing alone and 1-year post-walking infants. Sitting onset infants showed higher Vel_m and Vel_var than all other infants and lower F_m than walking onset and 1-year post-walking infants. 4. Discussion 4.1. Infants show adult-like patter of visual–postural coupling Our study is the first to show an inverted-U pattern for the frequency response of infants’ postural control to visual stimuli. From a few months post-sitting to one year post-walking, infants demonstrated higher gain responses at 0.52 and 0.76 Hz than at 0.12 and 1.24 Hz. Consistent with previous studies [11,12], we found higher gain response to 0.5–0.6 Hz than to 0.2–0.3 Hz in sitting infants. Furthermore, with a wider frequency range, we were able to reveal an adult-like inverted-U pattern of the visual– postural relationship in infants. While adults showed the strongest postural entrainment around 0.2 Hz in standing [6,7], the highest gain responses of infants in sitting appeared at higher frequencies. This discrepancy may result from different tasks (standing vs. sitting) and/or different eigenfrequency of postural sway in adults and infants which may be due to the differences in body height and/or the control system. Whether this frequency-dependent
160
L.-C. Chen et al. / Gait & Posture 45 (2016) 157–163
Fig. 2. Gain and phase from the averaged transfer functions at different amplitude conditions of SO (A, B) and W12 (C, D) infants to 0.52 Hz (A, C) and 1.24 Hz (B, D) stimuli. Values of gain are indicated on the vertical axis. Values of phase are indicated on the circle in units of degrees.
pattern changes later in childhood or differs by postural task requires further investigation. Human posture is temporally entrained to sensory information in similar ways in young adults, children, and infants [4–7,12– 14,19, present study]. From low to high frequency, all infants, except for new sitters, gradually changed their sitting postural sway from leading to lagging behind the visual motion with inphase entrainment around 0.28 Hz. Experienced walkers demonstrated a similar frequency-dependent temporal relationship of the visual–postural coupling across amplitude conditions. New sitters, however, showed no consistent response to the visual stimulus. Similar to previous studies that examined muscle activities [20], our results showed that new sitters exhibit more variable postural sway than experienced walkers. Their high level of self-motion may prevent precise detection and responses to the property (frequency) of the visual stimulus. Previous studies also found increases in the temporal consistency between infants’ postural sway and an oscillating somatosensory drive or a platform perturbation as they gained experience in walking [16,21]. Combining these results, we suggest that perception-action coupling of
postural behavior exists early in life but the developmental process involves fine-tuning the spatial and temporal properties of this coupling relationship. Walking plays an important role in the development of sensorimotor control of posture [16,21,22]. Conversely, development of the sensorimotor system may also play an important role in walking. 4.2. Infant posture re-weights to changes of visual amplitude Infants, except for new sitters, demonstrated evidence of sensory re-weighting in controlling their sitting posture. When the stimulus amplitude increased, infants down-weighted the visual information and showed a decreased gain response. These results, like those of adults [1,14] and children [3], suggest that the sensorimotor control of human posture is adaptive as early as a few months after sitting onset. A previous study showed that 3–6 months old infants were able to discriminate between optic flows only with large changes in heading [23]. Infants at 10–11 months of age were able to differentiate the speed of looming speed while 5–7-month olds processed all visual stimuli as fast looms [24]. In
L.-C. Chen et al. / Gait & Posture 45 (2016) 157–163
161
Fig. 3. Gain (A, C, E, G) and corresponding phase (B, D, F, H) (mean S.E.) of the postural responses to the visual stimulus across amplitude conditions and frequencies for infants at sitting onset (SO), at standing alone (ST), at walk onset (WO), and at 1-year post-walking (W12). Data are presented as mean + S.E.
the present study, when the visual stimulus was large, infants were less successful in maintaining quiet sitting and showed disruptive behaviors, such as turning to the parent or becoming fussy. Indeed, one new sitter was unable to remain in the sitting task during the large amplitude condition. Visual flow provides information about self-motion. The disruptive behavior in the large amplitude condition suggests that infants did perceive the visual information but were unable to successfully re-weight the visual information and thus disturbed by the conflict between perception and selfmotion. Our results explain previous findings that infants become less and less perturbed by optic flow and are better able to maintain equilibrium as upright postural experience increases [25–27].
Combining the findings in behavioral observation and visual– postural coupling, we suggest that perceptual abilities and control of the body both play important roles in infants’ postural behavior. Postural development involves a refinement of the sensorypostural relationship that allows the infant to respond appropriately to sensory information and to successfully re-weight the information when sensory condition changes. 4.3. Changes of sitting posture when walking emerges As expected, infants’ quiet sitting posture became more stable after a few months of sitting. Around the onset of walking, infants
162
L.-C. Chen et al. / Gait & Posture 45 (2016) 157–163
Fig. 4. Position (A) and velocity (B) variability (mean S.E.) of the residual postural sway, with the components of driving frequencies removed, across amplitude conditions for infants at sitting onset (SO), at standing alone (ST), at walk onset (WO), and at 1-year post-walking (W12). Data are presented as mean + S.D. Brackets indicate significant pair-wise comparisons.
became less stable and increased their postural sway in sitting. Such postural instability in WO infants was also revealed by the residual postural sway in the dynamic vision conditions. These results duplicate our previous results that infants showed a transient disruption in sitting posture when learning to walk, suggesting sensorimotor re-calibration as a new postural behavior emerges that allows the control system to explore adaptations to function in the environment [15,28]. Interestingly, the transient
disruption was not present in the frequency of infants’ sitting postural sway. The decrease in sway frequency during postural development has also been shown in our previous study on infants’ quiet stance and is thought to indicate a continuous process of sensorimotor refinement during early postural development [22]. This study employed a cross-sectional design and investigated experience-related differences of infants’ dynamic visual–postural relationship. Due to the limitation of cross-sectional design, future
Fig. 5. Position variability (A), mean velocity (B), velocity variability (C), and median frequency (D) (mean S.E.) of the postural sway for infants at sitting onset (SO), at standing alone (ST), at walk onset (WO), and at 1-year post-walking (W12). Data are presented as mean + S.D. Brackets indicate significant pair-wise comparisons.
L.-C. Chen et al. / Gait & Posture 45 (2016) 157–163
longitudinal research is necessary to confirm the developmental changes of infant visual–postural control indicated by our findings. 5. Conclusion Our results show that, from a few months post-sitting to 1-year post-walking, infants are able to adapt their postural behavior to an oscillating visual stimulus and demonstrate sensory reweighting ability. While the experienced sitters and walkers showed adult-like frequency- and amplitude-dependent features of their visual–postural relationship, newly sitting infants demonstrated highly variable postural sway and did not respond to the visual stimulus. We suggest that visual–postural coupling and sensory re-weighting are fundamental processes that are present after a few months post sitting. Early postural development involves a refinement of the sensorimotor dynamics that entails a complementary relationship between improved control of selfmotion and sensitivity to environmental motion. Acknowledgements We much appreciated the infants and their parents for their time and effort they gave for participating in this study. We also thank Dr. Tim Kiemel for his advice in data analyses and the undergraduate students who assisted in data collection and reduction. Conflict of interest statement The authors declare that there are no known conflicts of interest regarding the work described in the present manuscript. References [1] Oie KS, Kiemel T, Jeka JJ. Multisensory fusion: simultaneous re-weighting of vision and touch for the control of human posture. Brain Res Cogn Brain Res 2002;14:164–76. [2] Peterka RJ. Sensorimotor integration in human postural control. J Neurophysiol 2002;88:1097–118. [3] Bair W-N, Kiemel T, Jeka J, Clark J. Development of multisensory reweighting for posture control in children. Exp Brain Res 2007;183:435–46. [4] Barela JA, Jeka JJ, Clark JE. Postural control in children. Coupling to dynamic somatosensory information. Exp Brain Res 2003;150:434–42. [5] Schmuckler MA. Children’s postural sway in response to low- and highfrequency visual information for oscillation. J Exp Psychol Hum Percept Perform 1997;23:528–45. [6] Dijkstra TM, Schoner G, Giese MA, Gielen CC. Frequency dependency of the action–perception cycle for postural control in a moving visual environment: relative phase dynamics. Biol Cybern 1994;71:489–501.
163
[7] Jeka JJ, Oie KS, Kiemel T. Multisensory information for human postural control: integrating touch and vision. Exp Brain Res 2000;134:107–25. [8] Peterka RJ, Benolken MS. Role of somatosensory and vestibular cues in attenuating visually induced human postural sway. Exp Brain Res 1995;105: 101–10. [9] Horak FB, Macpherson JM, Rowell LB, Shepherd JT. Postural orientation and equilibrium. In: Rowell LB, Sheperd JT, editors. Handbook of physiology: a critical, comprehensive presentation of physiological knowledge and concepts. Oxford: Oxford University Press; 1996. p. 255–92. [10] Jouen F, Lepecq JC, Gapenne O, Bertenthal BI. Optic flow sensitivity in neonates. Infant Behav Dev 2000;23:271–84. [11] Bertenthal BI, Rose JL, Bai DL. Perception-action coupling in the development of visual control of posture. J Exp Psychol Hum Percept Perform 1997;23: 1631–43. [12] Barela JA, Godoi D, Freitas J, Polastri PF. Visual information and body sway coupling in infants during sitting acquisition. Infant Behav Dev 2000;23: 285–97. [13] Bertenthal BI, Boker SM, Xu M. Analysis of the perception–action cycle for visually induced postural sway in 9-month-old sitting infants. Infant Behav Dev 2000;23:299–315. [14] Kiemel T, Oie K, Jeka JJ. Slow dynamics of postural sway are in the feedback loop. J Neurophysiol 2006;95:1410–8. [15] Chen LC, Metcalfe JS, Jeka JJ, Clark JE. Two steps forward and one back: learning to walk affects infants’ sitting posture. Infant Behav Dev 2007;30: 16–25. [16] Metcalfe JS, McDowell K, Chang TY, Chen LC, Jeka JJ, Clark JE. Development of somatosensory-motor integration: an event-related analysis of infant posture in the first year of independent walking. Dev Psychobiol 2005;46:19–35. [17] Prieto TE, Myklebust JB, Hoffmann RG, Lovett EG, Myklebust BM. Measures of postural steadiness: differences between healthy young and elderly adults. IEEE Trans Biomed Eng 1996;43:956–66. [18] Metcalfe JS, Chen L-C, Chang T-Y, McDowell K, Jeka JJ, Clark JE. The temporal organization of posture changes during the first year of independent walking. Exp Brain Res 2005;161:405–16. [19] Jeka J, Oie K, Schoner G, Dijkstra T, Henson E. Position and velocity coupling of postural sway to somatosensory drive. J Neurophysiol 1998;79:1661–74. [20] Hadders-Algra M. Development of postural control during the first 18 months of life. Neural Plas 2005;12:99–108. [21] Cignetti F, Zedka M, Vaugoyeau M, Assaiante C. Independent walking as a major skill for the development of anticipatory postural control: evidence from adjustments to predictable perturbations. PLOS ONE 2013;8: e56313. [22] Chen LC, Metcalfe JS, Chang TY, Jeka JJ, Clark JE. The development of infant upright posture: sway less or sway differently. Exp Brain Res 2008;186: 293–303. [23] Gilmore RO, Baker TJ, Grobman KH. Stability in young infant’s discrimination of optic flow. Dev Psychol 2004;40:259–70. [24] Van der Weel FR, Van der Meer AL. Seeing it coming: infants’ brain responses to looming danger. Naturwissenschaften 2009;96:1385–91. [25] Butterworth G, Hicks L. Visual proprioception and postural stability in infancy. A developmental study. Perception 1977;6:255–62. [26] Forssberg H, Nashner LM. Ontogenetic development of postural control in man: adaptation to altered support and visual conditions during stance. J Neurosci 1982;2:545–52. [27] Stoffregen TA, Schmuckler MA, Gibson EJ. Use of central and peripheral optical flow in stance and locomotion in young walkers. Perception 1987;16:113–9. [28] Harbourne RT, Stergiou N. Nonlinear analysis of the development of sitting postural control. Dev Psychobiol 2003;42:368–77.
Contents lists available at ScienceDirect
Gait & Posture journal homepage: www.elsevier.com/locate/gaitpost
Development of adaptive sensorimotor control in infant sitting posture Li-Chiou Chen a,b,*, John Jeka c,d,e, Jane E. Clark c,d a
School and Graduate Institute of Physical Therapy, National Taiwan University, Taipei, Taiwan Physical Therapy Center, National Taiwan University Hospital, Taipei, Taiwan c Department of Kinesiology, University of Maryland, College Park, MD, USA d Graduate Program in Neuroscience and Cognitive Science, University of Maryland, College Park, MD, USA e Department of Kinesiology, Temple University, Philadelphia, PA, USA b
A R T I C L E I N F O
A B S T R A C T
Article history: Received 28 February 2015 Received in revised form 4 November 2015 Accepted 12 January 2016
A reliable and adaptive relationship between action and perception is necessary for postural control. Our understanding of how this adaptive sensorimotor control develops during infancy is very limited. This study examines the dynamic visual–postural relationship during early development. Twenty healthy infants were divided into 4 developmental groups (each n = 5): sitting onset, standing alone, walking onset, and 1-year post-walking. During the experiment, the infant sat independently in a virtual moving-room in which anterior-posterior oscillations of visual motion were presented using a sum-of-sines technique with five input frequencies (from 0.12 to 1.24 Hz). Infants were tested in five conditions that varied in the amplitude of visual motion (from 0 to 8.64 cm). Gain and phase responses of infants’ postural sway were analyzed. Our results showed that infants, from a few months post-sitting to 1 year post-walking, were able to control their sitting posture in response to various frequency and amplitude properties of the visual motion. Infants showed an adult-like inverted-U pattern for the frequency response to visual inputs with the highest gain at 0.52 and 0.76 Hz. As the visual motion amplitude increased, the gain response decreased. For the phase response, an adult-like frequency-dependent pattern was observed in all amplitude conditions for the experienced walkers. Newly sitting infants, however, showed variable postural behavior and did not systemically respond to the visual stimulus. Our results suggest that visual–postural entrainment and sensory re-weighting are fundamental processes that are present after a few months post sitting. Sensorimotor refinement during early postural development may result from the interactions of improved self-motion control and enhanced perceptual abilities. ß 2016 Elsevier B.V. All rights reserved.
Keywords: Posture Vision Sensorimotor Infant Sitting
1. Introduction Postural control is an important motor skill acquired during early development. To control the multi-segmented body in various conditions, a reliable and adaptive relationship between action and perception is necessary. Evidence of adaptive sensorimotor control in posture has been shown in adults [1,2] and in children [3–5]. Little is known about how this type of control develops during infancy. Young adults entrain their standing posture to the frequency properties of vision and somatosensory information [1,6,7]. The
* Corresponding author at: School and Graduate Institute of Physical Therapy, National Taiwan University, No. 17, Xuzhou Rd., Taipei 100, Taiwan. Tel.: +886 2 33668135; fax: +886 2 33668161. E-mail address: [email protected] (L.-C. Chen). http://dx.doi.org/10.1016/j.gaitpost.2016.01.020 0966-6362/ß 2016 Elsevier B.V. All rights reserved.
frequency response of this sensory-postural relationship demonstrates an inverted-U pattern with the greatest in-phase entrainment near 0.2 Hz and weaker and out-of-phase entrainment as the stimulus frequency decreases or increases [6,7]. Furthermore, adults’ postural sway is proportional to the stimulus amplitude within a range. When the amplitude exceeds a certain value, postural response decreases [2,8]. This change in the postural response to a change in the sensory stimuli indicates sensory re-weighting, a critical component of the adaptive sensorimotor control [1,2,9]. When a source of sensory information is unreliable, the postural system needs to attenuate this source of information and increase reliance on another modality. This sensory re-weighting process has been shown in children as young as 4 years of age [3] but has not been investigated in very young children or infants. Newborn infants show directionally appropriate and velocityscaled head response to visual flow information [10], suggesting
158
L.-C. Chen et al. / Gait & Posture 45 (2016) 157–163
that the visual–postural relationship may be a fundamental component of the sensorimotor system. A few studies have investigated infants’ postural responses to changes in the frequency of visual stimuli [11–13]. However, the results were conflicting. While some studies showed that infants’ postural entrainment increased as the frequency of the visual motion increased [11,12], one study reported a linear decline in sway coherence as a function of stimulus frequency [13]. Furthermore, the developmental changes in infants’ visual–postural coupling were inconsistent between the studies [11,12]. These conflicting results may be due to differences in the postural measures as well as the range of sensory properties employed in the studies. Since children were shown to respond to a wider frequency range of visual stimuli compared to adults [5], a sufficient range of sensory properties is required to better characterize the sensorimotor relationships in infants’ postural control. In the present study, we sought to systematically examine the dynamic visual–postural relationship in infants as they develop postural control. Specifically, we examined infants’ ability to adapt their sitting posture to different properties of the visual signal, i.e., frequency and amplitude. Incorporating a wide range of input frequency and amplitude, we addressed the following research questions: (1) does the adult patter (inverted-U) of visual–postural coupling exist in early infancy? (2) is sensory re-weighting a fundamental process of postural control present in young infants? Using a cross-sectional design, we also examined how the adaptive visual–postural relationship may differ by infants’ advances and experience in postural development from sitting to standing and finally to walking.
2. Method 2.1. Participants Twenty healthy infants (13 boys) composed four groups (each n = 5): (1) sitting onset (SO; aged 6.7 1.1 months; 6.0 5.1 days after sitting onset), (2) standing alone (ST; aged 10.6 1.2 months; 28.8 11.9 days before walking onset), (3) walking onset (WO; aged 11.7 1.4 months; 8.8 8.1 days after walking onset), and (4) 1-year post-walking (W12; aged 23.5 1.2 months; 11.7 1.0 months after walking onset). Sitting onset was when the infant could sit without support for 10 s. Standing alone was when the infant first able to stand without support for 5 s. Walking onset was when the infant started to walk independently for three continuous steps. Although no formal developmental assessment was done, the first author, a pediatric physical therapist, screened all infants to assure ageappropriate development. All experimental procedures were approved by the Institutional Review Board of the University. Parents gave written informed consent before participation.
Fig. 1. An infant sits independently on a chair in a 3-wall room. The smiley face represents the location of the cartoon video.
Infants’ postural sway was measured using an active infrared position tracking system (Optotrak, Northern Digital Inc.). Three small infrared LEDs were affixed to the infant’s occipital prominence, upper and middle trunk. A bank of three cameras was positioned parallel to the front screen and 2 m behind the infant. In addition, the testing session was videotaped for later coding of the infant’s sitting behavior. 2.3. Experimental design and procedures The visual stimulus was created by anterior–posterior (AP) oscillations of the animated display using a sum-of-sines technique [14] that allowed us to examine infants’ visual–postural coupling over a large range of frequency and amplitude variations without increasing the testing burden. The sum-of-sines consisted of a summation of 5 sinusoids at frequencies 0.12, 0.28, 0.52, 0.76, and 1.24 Hz with baseline amplitudes of 0.417, 0.179, 0.096, 0.065, and 0.040 cm, respectively. Five amplitude conditions were tested: A0: The visual display was stationary. A1: The visual display oscillated with the baseline amplitudes as described above. Peak-to-peak amplitude was 1.44 cm. A2: The amplitudes were twice of those in A1 condition. A4: The amplitudes were four times of those in A1 condition. A12: The amplitudes were twelve times of those in A1 condition. During the experiment, one experimenter and the parent stayed near the infant but not within his/her sight to provide help when needed. Data were collected in 3 randomized blocks, each with one 60-s trial for each amplitude condition.
2.2. Apparatus and measures 2.4. Data reduction and analysis Fig. 1 illustrates the experimental set-up in which the infant sat on a customized chair that was fixed on a 45 cm high pedestal and placed 100 cm from the front wall. The chair had a small back support and a safety belt loosely across the infant’s hip. The visual stimulus was created in a Fakespace Systems CAVE Automatic Virtual EnvironmentTM that is a rear-projected 3-screen display (each 2.5 m 3.0 m) with 1280 1024 pixels spatial resolution and 60 Hz framing rate. The visual display was specified by white triangles (0.28 0.38 0.28) randomly projected on a black background, excluding the foveal region (15 cm diameter circle) on the front screen. To attract the infant’s attention to the front screen, a cartoon video was projected onto the foveal region with auditory outputs behind the front screen.
Videos of all sitting trials were coded for usable time segments. In this study, the shortest period required to characterize all sum-of-sines components is 8.33 s (for the lowest frequency 0.12 Hz). For comparison with previous studies [15,16] and for coding convenience, we decided the minimal duration for sitting time segments as 10 s. The coding criterion was that the infant maintained quiet posture and continuous visual engagement to the front screen for at least 10 s. Only those time segments were used for subsequent data analysis. Mean segment time (MST) was computed across all segments of each infant in each amplitude condition to represent the infant’s engagement in the sitting task.
L.-C. Chen et al. / Gait & Posture 45 (2016) 157–163
Infants’ postural sway was analyzed only in the AP direction as it was the direction of visual stimulus. Head and trunk displacements were mean-detrended and low-pass filtered (recursive 2ndorder Butterworth filter, fcut-off = 5 Hz). For infants’ postural sway at the driving frequencies, the transfer (frequency-response) function at each frequency was computed dividing the Fourier transform of the postural response by that of the stimulus. Gain and phase were derived from the mean transfer function [7]. Gain represents the strength of the postural response to the stimulus. Larger gain values suggest more responsive to the stimulus. Phase is normalized time delay between the postural sway and the stimulus motion. A positive phase value indicates that the body movement leads the visual stimulus. Phase values that are not tightly clustered suggest inconsistent response to the stimulus. For infants’ postural responses at the non-driving frequencies, mean position variability (Pos_var) and velocity variability (Vel_var) were calculated for the residual postural sway, with the components of all 5 driving frequencies removed. To characterize infants’ quiet sitting posture, postural sway in the A0 condition was examined using distance-related and raterelated measures. Distance-related measures, including Pos_var, Vel_var, and mean velocity (Vel_m), were calculated to represent postural steadiness. Rate-related measures, including velocity and frequency variables, were used to indicate changes of postural control strategies [17,18]. For sway frequency, power spectrum density of postural sway time series was computed using multitaper method with 8 tapers and median frequency (F_m) was calculated. Mean for each postural measure was calculated, weighted by the segment duration. 2.5. Statistical analysis To examine infants’ postural responses to the moving visual stimulus, transfer functions were compared in the complex plane using a linear model of repeated measures MANOVA. The dependent variables were the real and imaginary parts and the independent variables were frequency, amplitude, and group. This analysis assessed the transfer function distributions among the independent variables and therefore took into account both gain and phase. Transfer functions with different gain responses but large phase variability may not be seen as different in this analysis. To further examine infants’ spatial and temporal postural responses to the visual stimulus, gain and phase were separately examined using mixed model repeated measures ANOVAs. To assess the effects of infant group and stimulus amplitude on the residual postural sway and to examine whether infants’ unperturbed sitting posture differed among the four groups, mixed model repeated measures ANOVAs were conducted. All statistical analyses were performed using the Statistical Analysis Software (SAS) program (Release 9.1, SAS Institute Inc.). The a level was set at 0.05 and post hoc comparisons with Tukey adjustment were performed when applicable. 3. Results All infants completed the experiment except that one sitting onset infant was unable to remain sitting in the A12 condition. MST results showed a significant group effect (p < 0.05). One-year postwalking infants engaged in the sitting task longer than all other infants (MST 36.5 8.9 and 25.8 15.6 s, respectively). No significant difference was found among amplitude conditions or the three younger groups. Infants’ postural responses at head, upper and middle trunk showed similar patterns among the groups, amplitude conditions, and frequencies. As would be expected for segments more distal to the support surface, the postural response was larger for the
159
head than for the trunk. Therefore, our current analysis focused on results from head data. 3.1. Postural responses to the visual stimulus: frequency and amplitude responses The transfer functions for infants’ postural responses to the visual stimulus at the driving frequencies did not cluster on the complex plane (see exemplars in Fig. 2). Repeated measures MANOVA revealed a significant frequency effect (Wilks’ Lambda = 0.79, p < 0.05) and a marginal group frequency effect (Wilks’ Lambda = 0.89, p = 0.064). Significant frequency effects were found in all infant groups except for sitting onset infants. Gain and phase of infants’ postural responses to the visual stimulus were further separately analyzed. For gain, significant effects of frequency, amplitude, group, and group amplitude were found (all p < 0.001). All infants showed significantly higher gain responses at 0.52 and 0.76 Hz than at 0.12 and 1.24 Hz. As the stimulus amplitude increased, gain response decreased (Fig. 3). A significant group effect existed only in the A1 condition in which sitting onset infants showed higher gains than all other infants. For phase, significant frequency, frequency group, frequency amplitude, and frequency amplitude group effects were found (all p < 0.05). All infants demonstrated positive phase (posture leading stimulus) at lower frequencies and gradually became negative phase (lagging behind) at higher frequencies. One-year post-walking infants showed significant phase differences between 0.52 and 1.24 Hz in all amplitude conditions while sitting onset infants demonstrated no frequency response in any amplitude condition. For infants’ residual postural sway at the non-driving frequencies, a significant group effect was revealed for both Pos_var and Vel_var (p < 0.001). Sitting onset infants demonstrated significantly higher Pos_var than standing alone and 1-year post-walking infants, and walking onset infants higher than 1-year post-walking infants. Vel_var was highest in sitting onset infant and lowest in 1year post-walking infants (Fig. 4). 3.2. Unperturbed sitting posture For infants’ unperturbed sitting posture, all sway measures in the A0 condition showed significant group effects (all p < 0.05; Fig. 5). Pos_var was larger in sitting onset infants than in standing alone and 1-year post-walking infants. Sitting onset infants showed higher Vel_m and Vel_var than all other infants and lower F_m than walking onset and 1-year post-walking infants. 4. Discussion 4.1. Infants show adult-like patter of visual–postural coupling Our study is the first to show an inverted-U pattern for the frequency response of infants’ postural control to visual stimuli. From a few months post-sitting to one year post-walking, infants demonstrated higher gain responses at 0.52 and 0.76 Hz than at 0.12 and 1.24 Hz. Consistent with previous studies [11,12], we found higher gain response to 0.5–0.6 Hz than to 0.2–0.3 Hz in sitting infants. Furthermore, with a wider frequency range, we were able to reveal an adult-like inverted-U pattern of the visual– postural relationship in infants. While adults showed the strongest postural entrainment around 0.2 Hz in standing [6,7], the highest gain responses of infants in sitting appeared at higher frequencies. This discrepancy may result from different tasks (standing vs. sitting) and/or different eigenfrequency of postural sway in adults and infants which may be due to the differences in body height and/or the control system. Whether this frequency-dependent
160
L.-C. Chen et al. / Gait & Posture 45 (2016) 157–163
Fig. 2. Gain and phase from the averaged transfer functions at different amplitude conditions of SO (A, B) and W12 (C, D) infants to 0.52 Hz (A, C) and 1.24 Hz (B, D) stimuli. Values of gain are indicated on the vertical axis. Values of phase are indicated on the circle in units of degrees.
pattern changes later in childhood or differs by postural task requires further investigation. Human posture is temporally entrained to sensory information in similar ways in young adults, children, and infants [4–7,12– 14,19, present study]. From low to high frequency, all infants, except for new sitters, gradually changed their sitting postural sway from leading to lagging behind the visual motion with inphase entrainment around 0.28 Hz. Experienced walkers demonstrated a similar frequency-dependent temporal relationship of the visual–postural coupling across amplitude conditions. New sitters, however, showed no consistent response to the visual stimulus. Similar to previous studies that examined muscle activities [20], our results showed that new sitters exhibit more variable postural sway than experienced walkers. Their high level of self-motion may prevent precise detection and responses to the property (frequency) of the visual stimulus. Previous studies also found increases in the temporal consistency between infants’ postural sway and an oscillating somatosensory drive or a platform perturbation as they gained experience in walking [16,21]. Combining these results, we suggest that perception-action coupling of
postural behavior exists early in life but the developmental process involves fine-tuning the spatial and temporal properties of this coupling relationship. Walking plays an important role in the development of sensorimotor control of posture [16,21,22]. Conversely, development of the sensorimotor system may also play an important role in walking. 4.2. Infant posture re-weights to changes of visual amplitude Infants, except for new sitters, demonstrated evidence of sensory re-weighting in controlling their sitting posture. When the stimulus amplitude increased, infants down-weighted the visual information and showed a decreased gain response. These results, like those of adults [1,14] and children [3], suggest that the sensorimotor control of human posture is adaptive as early as a few months after sitting onset. A previous study showed that 3–6 months old infants were able to discriminate between optic flows only with large changes in heading [23]. Infants at 10–11 months of age were able to differentiate the speed of looming speed while 5–7-month olds processed all visual stimuli as fast looms [24]. In
L.-C. Chen et al. / Gait & Posture 45 (2016) 157–163
161
Fig. 3. Gain (A, C, E, G) and corresponding phase (B, D, F, H) (mean S.E.) of the postural responses to the visual stimulus across amplitude conditions and frequencies for infants at sitting onset (SO), at standing alone (ST), at walk onset (WO), and at 1-year post-walking (W12). Data are presented as mean + S.E.
the present study, when the visual stimulus was large, infants were less successful in maintaining quiet sitting and showed disruptive behaviors, such as turning to the parent or becoming fussy. Indeed, one new sitter was unable to remain in the sitting task during the large amplitude condition. Visual flow provides information about self-motion. The disruptive behavior in the large amplitude condition suggests that infants did perceive the visual information but were unable to successfully re-weight the visual information and thus disturbed by the conflict between perception and selfmotion. Our results explain previous findings that infants become less and less perturbed by optic flow and are better able to maintain equilibrium as upright postural experience increases [25–27].
Combining the findings in behavioral observation and visual– postural coupling, we suggest that perceptual abilities and control of the body both play important roles in infants’ postural behavior. Postural development involves a refinement of the sensorypostural relationship that allows the infant to respond appropriately to sensory information and to successfully re-weight the information when sensory condition changes. 4.3. Changes of sitting posture when walking emerges As expected, infants’ quiet sitting posture became more stable after a few months of sitting. Around the onset of walking, infants
162
L.-C. Chen et al. / Gait & Posture 45 (2016) 157–163
Fig. 4. Position (A) and velocity (B) variability (mean S.E.) of the residual postural sway, with the components of driving frequencies removed, across amplitude conditions for infants at sitting onset (SO), at standing alone (ST), at walk onset (WO), and at 1-year post-walking (W12). Data are presented as mean + S.D. Brackets indicate significant pair-wise comparisons.
became less stable and increased their postural sway in sitting. Such postural instability in WO infants was also revealed by the residual postural sway in the dynamic vision conditions. These results duplicate our previous results that infants showed a transient disruption in sitting posture when learning to walk, suggesting sensorimotor re-calibration as a new postural behavior emerges that allows the control system to explore adaptations to function in the environment [15,28]. Interestingly, the transient
disruption was not present in the frequency of infants’ sitting postural sway. The decrease in sway frequency during postural development has also been shown in our previous study on infants’ quiet stance and is thought to indicate a continuous process of sensorimotor refinement during early postural development [22]. This study employed a cross-sectional design and investigated experience-related differences of infants’ dynamic visual–postural relationship. Due to the limitation of cross-sectional design, future
Fig. 5. Position variability (A), mean velocity (B), velocity variability (C), and median frequency (D) (mean S.E.) of the postural sway for infants at sitting onset (SO), at standing alone (ST), at walk onset (WO), and at 1-year post-walking (W12). Data are presented as mean + S.D. Brackets indicate significant pair-wise comparisons.
L.-C. Chen et al. / Gait & Posture 45 (2016) 157–163
longitudinal research is necessary to confirm the developmental changes of infant visual–postural control indicated by our findings. 5. Conclusion Our results show that, from a few months post-sitting to 1-year post-walking, infants are able to adapt their postural behavior to an oscillating visual stimulus and demonstrate sensory reweighting ability. While the experienced sitters and walkers showed adult-like frequency- and amplitude-dependent features of their visual–postural relationship, newly sitting infants demonstrated highly variable postural sway and did not respond to the visual stimulus. We suggest that visual–postural coupling and sensory re-weighting are fundamental processes that are present after a few months post sitting. Early postural development involves a refinement of the sensorimotor dynamics that entails a complementary relationship between improved control of selfmotion and sensitivity to environmental motion. Acknowledgements We much appreciated the infants and their parents for their time and effort they gave for participating in this study. We also thank Dr. Tim Kiemel for his advice in data analyses and the undergraduate students who assisted in data collection and reduction. Conflict of interest statement The authors declare that there are no known conflicts of interest regarding the work described in the present manuscript. References [1] Oie KS, Kiemel T, Jeka JJ. Multisensory fusion: simultaneous re-weighting of vision and touch for the control of human posture. Brain Res Cogn Brain Res 2002;14:164–76. [2] Peterka RJ. Sensorimotor integration in human postural control. J Neurophysiol 2002;88:1097–118. [3] Bair W-N, Kiemel T, Jeka J, Clark J. Development of multisensory reweighting for posture control in children. Exp Brain Res 2007;183:435–46. [4] Barela JA, Jeka JJ, Clark JE. Postural control in children. Coupling to dynamic somatosensory information. Exp Brain Res 2003;150:434–42. [5] Schmuckler MA. Children’s postural sway in response to low- and highfrequency visual information for oscillation. J Exp Psychol Hum Percept Perform 1997;23:528–45. [6] Dijkstra TM, Schoner G, Giese MA, Gielen CC. Frequency dependency of the action–perception cycle for postural control in a moving visual environment: relative phase dynamics. Biol Cybern 1994;71:489–501.
163
[7] Jeka JJ, Oie KS, Kiemel T. Multisensory information for human postural control: integrating touch and vision. Exp Brain Res 2000;134:107–25. [8] Peterka RJ, Benolken MS. Role of somatosensory and vestibular cues in attenuating visually induced human postural sway. Exp Brain Res 1995;105: 101–10. [9] Horak FB, Macpherson JM, Rowell LB, Shepherd JT. Postural orientation and equilibrium. In: Rowell LB, Sheperd JT, editors. Handbook of physiology: a critical, comprehensive presentation of physiological knowledge and concepts. Oxford: Oxford University Press; 1996. p. 255–92. [10] Jouen F, Lepecq JC, Gapenne O, Bertenthal BI. Optic flow sensitivity in neonates. Infant Behav Dev 2000;23:271–84. [11] Bertenthal BI, Rose JL, Bai DL. Perception-action coupling in the development of visual control of posture. J Exp Psychol Hum Percept Perform 1997;23: 1631–43. [12] Barela JA, Godoi D, Freitas J, Polastri PF. Visual information and body sway coupling in infants during sitting acquisition. Infant Behav Dev 2000;23: 285–97. [13] Bertenthal BI, Boker SM, Xu M. Analysis of the perception–action cycle for visually induced postural sway in 9-month-old sitting infants. Infant Behav Dev 2000;23:299–315. [14] Kiemel T, Oie K, Jeka JJ. Slow dynamics of postural sway are in the feedback loop. J Neurophysiol 2006;95:1410–8. [15] Chen LC, Metcalfe JS, Jeka JJ, Clark JE. Two steps forward and one back: learning to walk affects infants’ sitting posture. Infant Behav Dev 2007;30: 16–25. [16] Metcalfe JS, McDowell K, Chang TY, Chen LC, Jeka JJ, Clark JE. Development of somatosensory-motor integration: an event-related analysis of infant posture in the first year of independent walking. Dev Psychobiol 2005;46:19–35. [17] Prieto TE, Myklebust JB, Hoffmann RG, Lovett EG, Myklebust BM. Measures of postural steadiness: differences between healthy young and elderly adults. IEEE Trans Biomed Eng 1996;43:956–66. [18] Metcalfe JS, Chen L-C, Chang T-Y, McDowell K, Jeka JJ, Clark JE. The temporal organization of posture changes during the first year of independent walking. Exp Brain Res 2005;161:405–16. [19] Jeka J, Oie K, Schoner G, Dijkstra T, Henson E. Position and velocity coupling of postural sway to somatosensory drive. J Neurophysiol 1998;79:1661–74. [20] Hadders-Algra M. Development of postural control during the first 18 months of life. Neural Plas 2005;12:99–108. [21] Cignetti F, Zedka M, Vaugoyeau M, Assaiante C. Independent walking as a major skill for the development of anticipatory postural control: evidence from adjustments to predictable perturbations. PLOS ONE 2013;8: e56313. [22] Chen LC, Metcalfe JS, Chang TY, Jeka JJ, Clark JE. The development of infant upright posture: sway less or sway differently. Exp Brain Res 2008;186: 293–303. [23] Gilmore RO, Baker TJ, Grobman KH. Stability in young infant’s discrimination of optic flow. Dev Psychol 2004;40:259–70. [24] Van der Weel FR, Van der Meer AL. Seeing it coming: infants’ brain responses to looming danger. Naturwissenschaften 2009;96:1385–91. [25] Butterworth G, Hicks L. Visual proprioception and postural stability in infancy. A developmental study. Perception 1977;6:255–62. [26] Forssberg H, Nashner LM. Ontogenetic development of postural control in man: adaptation to altered support and visual conditions during stance. J Neurosci 1982;2:545–52. [27] Stoffregen TA, Schmuckler MA, Gibson EJ. Use of central and peripheral optical flow in stance and locomotion in young walkers. Perception 1987;16:113–9. [28] Harbourne RT, Stergiou N. Nonlinear analysis of the development of sitting postural control. Dev Psychobiol 2003;42:368–77.