Animal models offer the opportunity to acquire a new perspective on motor development.

Animal Models Offer the Opportunity to Acquire a New Perspective on Motor Development When students are introduced to the area of sensorimotor development in pediatrics, they are commonly directed to literature that describes the acquisition of new skills and the clinical tools used to measure the level of skills acquired by an infant or child. It may also be suggested that behavioral gains during development are directly attributed to a corresponding stage of anatomical or physiological maturation. Terms such a "cephalocaudal progression," "myelinization," "encephalization," "developmental reflexes," and "developmental sequences" quickly become a part of the student's vocabulary and the framework for conceptualizing (hypothetical) developmental processes. During much of this century, such conceptualizations dramatically shaped ideas regarding the process of development and the treatment of developmental disabilities. Although traditional ideas regarding development and clinical approaches to treatment may continue to be useful in the practice of pediatrics, recent basic science studies of sensorimotor development in a variety of animal species offer new perspectives regarding the processes that result in behavioral maturation. Furthermore, basic science studies of development, using animal models, offer the potential to design and test both old and clinical treatment procedures with a scientific rigor that is usually dificult to achieve in clinical studies (for further discussion of this issue, see Bradley and Bekoff [1] ).

In this article, I will first briefly review some of the historical notions concerning early sensorimotor development and some of the more recent literature that has modified my own perspective. The methods section will describe the procedures used to study motor development in neonatal kittens and chick embryos. The results section will provide a summary of findings from several of the studies. In the discussion section, a case will be made for what appear to be some important developmental issues raised by the studies. Possible implications of recent studies will be considered with respect to the practical concerns of researchers and clinicians seeking to establish a theoretical foundation forpediatric rehabilitation.

Historical Perspective:

Reflex-based Behavior

Nearly all infant assessment tools, such as those proposed by Prechtl and Beitema, [2] Dubowitz and Dubowitz, [3] and Brazelton, [4] include procedures for evaluating developmental reflexes. Historically, commonly held views of developmental reflexes were greatly influenced by the English physiologist CS Sherrington. Early in this century, Sherrington [5] proposed that motor behavior was reflex-based and that the stretch reflex was the basic building block for complex behavior (for a review, see chapter by Grillner [6]). thus, in this scientific climate, scientists set about examining various reflex behaviors in fetal animals, most notably the kitten, [7,8] and aborted human fetuses. [9,10] Sherrington's notion of reflex hierarchies was eventually ext ended to suggest that reflexes emerging during development are assembled hierarchically and reflect a progessive caudal-to-rostral shift in the neurological control of behavior. Apparently, it is the notion of reflex hierarchies that has inspired efforts to document the ages at which different reflexes can first be elicited. [11,12]

Developmental reflexes came to be viewed as means for determining the state of both normal and abnormal neural development. Considerable effort has been (and continues to be) invested in attempts to determine the possible relationship between the presence or absence of various reflexes and long-term developmental outcome. [13,14] The emergence of reflexes in late fetal development and the subsequent absence of these reflexes in the postnatal period were hypothesized to be indicative of an initial control of behavior by lower-order neural structures and subsequent overriding of those structures by higher-order neural structures (eg, encephalization). Perhaps more importantly, focus on these reflexive behaviors appears to have contributed to the view that the developing animal (including the human) is a primitive organism that lacks the capacity for adaptive behavior and the ability to initiate strategies for coping with its surroundings. A comprehensive discussion of these issues is provided by Touwen. [15]

Recent studies have extended our knowledge of developmental neurobiology beyond concepts of reflex control of movement. We now know that animals, including humans, have the capacity to exhibit an extensive repertoire of complex and adaptive behaviors much earlier in development than previously thought. Further, these studies suggest that if critical elements in behavioral maturation are to be understood and effectively manipulated, it will be important for both researchers and clinicians in pediatric rehabilitation to gain an understanding of a broad range of developmental processes that extent beyond changes in neural cell structure and function.

Some Recent Research

Developments

A rapid expansion in research efforts to study complex movement began in the early 1970s. Many of these studies focused on movements that are characterized by repetitive stereotypic features such as the electromyography (EMG) and biomechanics of leg motions during the step cycle for locomotion and the swipe cycle for grooming behaviors. The EMG, kinematics, and, in a few instances, kinematics, have been extensively studied and characterized for these repetitive movements in both humans [16,17] rats, [22] animals including cats, [18-21] rats, [22] frogs, [23] turtles, [24] and chicks. [25,26] Studies such as these described the fundamental patterns of neural output to muscles during ongoing behavior by analyzing the relative phasing of EMG bursts for agonist and antagonist muscles. In a number of animal studies, using curare to eliminate movement and L-dopa or an electrical stimulus to initiate neural activity (commonly referred to as a fictive preparation), many aspects of these patterns were also identified in neurogram, extracellular, and intracellular recordings at several regions within the nervous system. [6,18,27-30] Such studies led to the discovery that in the absence of both descending neural input from the brain and afferent feedback from the periphery, spinal neural networks can organize motoneuron pools to produce the fundamental patterns of activity corresponding to EMG recordings during limb movements for locomotion and grooming. [6] Knowledge that isolated spinal networks can produce complex movements requires not only that we reconsider previously assigned roles for reflex activity and higher-order motor centers, but also that we update our perspectives regarding the integrative nature of sensorimotor processes producing coordinated and impaired movement.

The extension of these discoveries to embryonic and neonatal animals suggests that we should also reassess our views regarding the acquisition of skilled movement during development. A number of developmental studies indicate that neural components for complex movements are established during the earliest phases of morphological development. For example, Stehouwer and Farel [30] reported that primary flexor and extensor motoneurons in the lumbar spinal cord of the frog tadpole were rhythmically and alternately active, as during locomotion, before the corresponding muscles differentiated. Bekoff [31] found that ankle flexor and extensor muscles in the chick embryo were already alternately active, as during locomotion, a day after onset of limb movements and 3 days prior to completion of the spinal reflex arc. furthermore, synchronized flexions and extensions of the hip, knee, and ankle, similar to those of locomotion, were observed within 2 to 3 days after the reported patterned activity in ankle antagonist muscle. [32] Although lower-order animals, such as the frog and chick, are phylogenetically distant from the human, the relevance of these studies is suggested by recent reports of coordinated limb excursions in ultrasound recordings of human fetuses. Human fetuses were reported to initiate alternating flexions and extensions unilaterally [33] and alternating leg movements bilaterally during postural changes ("somersaults") in utero [34] as early as 13 to 14 weeks gestational age. Wheter leg movements in human fetuses precede or emerge with functional lumbar reflexes is still somewhat unclear (for further discussion, see de Vries et al [34]). Nonetheless, it appears that human fetuses, like other animals, possess sufficient neural substrate to generate stepping-like limb movements at the earliest stages of development.

Recent studies also indicate that expression of sensorimotor abilities during development is highly context dependent and that very often an animal can initiate a complex movement much earlier than typically observed if certain variables are present. For example, birds do not initiate efforts to fly before the first week after hatching; however, when dropped a short distance (1 m}, they initiated bilaterally coordinated wing flapping within a day after hatching. [35] thus, the potential to perform coordinated wing movements is present in advance of the sill to fly, but stimuli attributable to gravity-enduced acceleration appear to be required to express this ability. Similarly, frog tadpoles initiated coordinated hindlimb movements far earlier in development when placed on a textured surface rather than in a tank of water. [23] Apparently, the ground reaction forces that emerged as the tadpoles moved over the surface were required to generate coordinated limb movements. Whereas, in a study of fetal and neonatal rats, coordinated (quadrupedal) limb movements were initiated several days earlier if the rat pups were placed in water rather than on a firm surface. [36] Additionally, if fetal or neonatal rats were placed in a postural support (chairlike) apparatus, they initiated complex grooming behaviors with their forelimbs a week earlier than typically obserbed. [37] In these later instances, it appears that a reduction in postural control requirements permitted an earlier expression of coordinated limb movements.

The studies discussed above appear to suggest that the potential to produce complex coordinated movement emerges very early in development and that many coordinated movements are components of skills that will be eventually acquired. Analogies can be found in recent human studies. for instance, when placed supine, infants do not typically reach for a dangling ring until the fourth postnatal month. [38] However, if supported in a seated posture, infants will initiate efforts to intercept a target moving across the visual field within the first postnatal month. [39] A salient moving target, an upright posture, and reduced postural requirements may be identified as possible context-dependent elements required to observe the expression of the infant's potential to produce a complex and adaptive behavior during early postnatal development. Taken altogether, recent studies of complex behavior suggest that, although developmental milestones are useful to distinguish levels of functional achievements, such milestones alone do not tell us very much about the early potential to produce complex movements or the processes that underly the eventual achievement of these milestones.

Finally, contrary to long-held notions that developmental reflexes are eventually inhibited for normal development to ensue, evidence also suggests that early forms of coordinated movement need not be extinguished with the subsequent differentiation and function of brain structures. For example, in chicks, prehatching movements are first initiated 2 to 3 days before hatching and consist of synchronous, repetitive flexions and extension of both legs so that the beak is thrust agaisnt the inner wall of the egg shell to break through it. Once the chick emerges from the shell, hatching behavior is no longer observed under free-ranging conditions. However, if at later ages a normal healthy chick is placed inside an artificial glass egg that accommodates the increase in body size, it will again initiate hatching behavior even as a young adult. [40] Thus, animals apparently retain the potential to exhibit some behaviors unique to a developmental phase and can again initiate these behavior if certain contextual elements are present to trigger or sustain them. Furthermore, retention of the potential to perform these behaviors need not compromise subsequent sensorimotor development. Similar arguments can be formed regarding the fate of the infant stepping reflex in humans. [41,42]

Method Used to Study

Neonatal and Embryonic

Motor Behavior

My own research efforts have focused on the study of motor development in kittens from birth to 2 postnatal months of age [43-45] and in chicks at embryonic days 9 and 10 (E9 and E10). [46,47] To study their motor development, I focused on the emergence of behaviors characterized by repetitive limb movements for locomotion, scratching, paw shaking, and embryonic motility using EMG and video recording methods. By selecting to study the development of repetitive limb movements, it was possible to use quantitative analysis procedures already established in the literature to evaluate the data and to compare results with the EMG and kinematic patterns for these behaviors in adult animals. Also, by selecting to study repetitive limb movements, I was able to interpret the results in light of numerous studies that have examined the physiological and biomechanical control of these behaviors in adult animals. Further, all of these behaviors, except for scratching, can be elicited in the hindlimbs of animals following complete transection of the thoracolumbar spinal cord. [48-50] Thus, to explore the role of higher-order neural inputs in motor development, emergence of these behaviors was also studied in neonatal and embryonic animals following spinal transection early in the postnatal (kittens) or embryonic (chicks) period. The methods used to study these behaviors are briefly summarized in the following sections. To critically examine the methods used, the reader is referred to the original, peer-reviewed studies.

Studies of Kitten Motor

Development

Surgical procedures. Neonatal kittens were anesthetized 24 hours after birth to surgically implant (under aseptic conditions) bipolar EMG electrodes. By surgically implanting the electrodes, it was possible to maintain the original electrodes for serial recordings over an extended period (2 months). To assure electrode position over multiple recordings, the electrode leads were sutured at several places between the muscle implanted and the hip, where the leads exited and inserted into a plug attachment. To allow for growth, the electrode leads were made extra long and portions of the leads were looped between each of the suture sites. A total of four electrodes were implanted in each kitten: three were implanted in the right leg to describe intralimb coordination, and one electrode was implanted in the left leg to describe interlimb coordination. Some of the kittens also received a spinal transection during the surgery that involved exposing and cutting the cord at the T-12 level. (In keeping with the original studies, kittens with an intact spinal cord will be referred to in the remaining text as "normal kittens" and kittens with a spinal transection will be referred to as "spinal kittens.")

Recording procedures. In an effort to obtain an optimal performance, kittens were tested in the presence of their mother during periods of nursing until they began to actively explore their surroundings (typically in the third postnatal week). To test for locomotor skills, EMG and synchronized video recordings were obtained as the kitten progressed unrestrained overground on a carpet and while supported with the hindlimbs pendent (airstepping) or in contact with the moving belt of a treadmill (bipedal treadmill locomotion). During the second postnatal month, normal kittens were also tested for quadrupedal treadmill locomotion. To test for scratching, the kitten was placed on the carpet and the pinna ipsilateral to the implanted hindlimb was stimulated with a dry or wet cotton swab or with a small ball of masking tape placed just inside the ear canal. To test for paw shaking, the kitten was again placed on the carpet and a small piece of masking tape was placed on the paw of the implanted hindlimb. Recordings were obtained two to three times a week for the first 2 postnatal months.

Analysis procedures. As stated previously, EMG and kinematic recordings of repetitive behaviors contain features that can be used to obtain reliable quantitative measurements. Typical of studies that measure EMG during repetitive limb movements, the ankle extensor (ie, lateral gastrocnemius [LG]) was selected as a reference muscle; the onset of consecutive LG bursts defined the onset and duration of a cycle period corresponding to one complete cycle of extension and flexion (Fig. 1). Burst onsets for other muscles active during that cycle were measured with respect to the LG burst onset to calculate an onset latency (absolute time) and a relative onset latency (percentage of cycle). The time from onset to offset for each burst was also measured to calculate absolute and relative burst durations. Finally, the synchronized video recordings were reviewed to describe contextual elements (eg, behavioral, environmental) in the EMG records analyzed and to help interpret the results.

Studies of Chick Motor

Development

Surgical procedures. To study motor development in the chick embryo, eggs were removed from the incubator on E9 or E10 and placed in a temperature-controlled bath. A window was made in the eggshell to expose the embryo lying on its left side. Working with the assistance of a stereomicroscope, blood vessels, membranes, and skin overlying the muscles selected for recording were carefully deflected to place bipolar electrodes perpendicular to the surface of each muscle. To obtain recordings during unrestricted movement in ovo, the electrodes were made of a very fine and flexible plastic tubing that could be held in place by light suction. In some instances, embryos previously received a low-thoracic spinal transection on E2, before differentiation of the spinal cord is complete and differentiation of limb buds has begun. To produce a spinal transection, three segmental levels of neural tube were removed under stereomicroscopic magnification. The egg was then resealed and returned to the incubator until E10 for EMG recording. (In keeping with the original studies, embryos with an intact spinal cord will be referred to in the remaining text as normal embryos and embryos with a spinal transection will be referred to as spinal embryos.)

Recording procedures. Normally, chick embryos are spontaneously and periodically active (referred to as spontaneous motility). There is no apparent external trigger for the activity, which involves movements of all body parts, and the spontaneous episodes typically occur every 2 to 3 minutes. Previous work demonstrated that these episodes of motility are characterized by repetitive cycles of muscle activity at the ankle. [31] Thus, to describe intralimb coordination in a given embryo, one pair of muscles with antagonist actions (flexion versus extension) at the ankle and a second pair with antagonist actions at the knee or hip were selected for recording. This recording configuration also provided information to characterize the extent of coordination between muscles with synergist actions (flexor and extensor synergies) at the ankle and knee or ankle and hip. Once electrodes were in place, EMG recordings ran uninterrupted for 1 hour, unless an electrode fell off and had to be replaced, any change was noted in pulse rate of the umbilical artery, or a change was noted in the periodicity of spontaneous activity.

Analysis procedures. The EMG data were analyzed using procedures similar to those described for the kitten studies. In the embryo studies, synchronized video methods were not applied to the microscopic setup because of technical limitations. Thus, to obtain some estimate of the behavioral context associated with the EMG recordings selected for analysis, verbal comments recorded on tape with the EMG data and written comments were reviewed. Embryos having received a spinal transection at E2 were placed in a tissue fixative at the end of the recording, and the spinal cord was subsequently stained for neural filaments to verify the extent of transection at the time of recording.

Summary of Results from

Neonatal and Embryonic

Studies

Development of Repetitive

Behaviors in Kittens

Adultlike EMG patterns were apparent in the earliest postnatal recordings of each behavior studied. When suckling was interrupted and neonatal kittens were placed a short distance from their mother, they rose onto all four limbs and took several awkward weight-supported steps to return to her side, frequently losing their balance in the process. [43] Limb movements were hypermetric, and there was little apparent coordination among the four limbs. Similarly, hypermetric limb movements were noted during stepping when the limbs were pendent (airstepping) and in contact with the treadmill belt. During these limb movements, coordination between the hindlimbs varied from synchronous to alternating. Yet, despite the visible immaturity of these limb movements, there were instances in each of the stepping conditions when several alternating steps were taken and accompanied by an EMG pattern that appeared similar to the adult pattern for locomotion. As illustrated in Figures 2A and 3A, the extensor muscles of the ankle (LG) and knee (vastus lateralis [VL]) were coactive for the majority of a given cycle period and reciprocally active with a brief burst (burst #2, Fig. 2A) in the ankle flexor muscle (tibialis anterior [TA]). During initial postnatal development, low-level TA activity (burst #1, Fig. 2A) also occurred co-actively with the extensor portion of the cycle period for overground and treadmill stepping, but was minimal during airstepping (Fig. 3A) obtained in the same recording sessions. As in adult records, extensor burst duration varied closely with cycle period, whereas the duration of the reciprocal flexor burst did not (Fig. 4).

Adultlike EMG patterns were also apparent in the earliest postnatal efforts to scratch (at 3 postnatal weeks) and paw shake (at 4 postnatal weeks). [45] During initial efforts to scratch, the kitten lowered its trunk to a nearly prone position or partially sitting position, shook its head vigorously, then flexed the ipsilateral hip to position the hindpaw adjacent to the ear during repetitive flexion and extension of the ankle and/or knee. Similarly, during initial paw-shaking responses, the kitten squatted on all four limbs, then slightly abducted the ipsilateral hip so that the paw cleared the floor during repeated flexion and extension of the entire leg. During both responses, trunk motions were of considerable amplitude, rocking side to side as the paw swiped the ear during scratching, and rocking rostrocaudally as the kitten attempted to shake the tape off the paw. Thus, during these responses, the kitten occasionally lost its balance. Yet, despite postural difficulties, the EMG recordings for the very first responses were nearly identical to responses at 2 postnatal months and to responses obtained in adult cats. [21] Scratching was characterized by reciprocal activation of the extensors (LG and VL) and ankle flexor (TA) at a frequency of 5 to 6 Hz, with flexor and extensor bursts similar in duration (Fig. 5). As in the adult cat, [49] paw shaking was characterized by the reciprocal activation of the ankle extensor (LG) with the mixed synergy (TA + VL) at a frequency of 10 Hz (Fig. 6A). Also characteristic of the adult pattern, the LG burst was brief, whereas the TA and VL bursts were longer in duration.

Maturation was characterized by greater ease and versatility of expression. Although maturation was not characterized by changes in the muscle patterns, several behavioral changes were noted. All of the behaviors, except airstepping, were more readily elicited with increasing age; airstepping was not observed after the second postnatal week. [43] Locomotion and paw-shaking responses were sustained for longer periods (eg, responses contained more continuous cycles), whereas scratching decreased (eg, responses contained fewer continuous cycles). [45] Cycle periods for overground and treadmill locomotion decreased in duration to adult values by 6 to 8 postnatal weeks. Along with these changes, quadrupedal posture became more erect and secure. The kittens also began to interrupt one behavior and interpose additional behaviors. For instance, in the first postnatal month, the kittens typically squatted and scratched nonstop for 20 or more cycles; in the second postnatal month, the kittens also initiated scratching from a seated or erect tripod stance and repeatedly interrupted the response after only a few cycles to play or explore. Thus, over the 2 postnatal months, the context for each behavior became more varied and the behaviors became more adaptive. That is, the duration of a behavior could be modified (eg, the kitten could locomote for a longer period), the postural context could be varied (eg, the kitten could initiate scratching from more than one position), and switching between behaviors could be more readily executed (eg, the kitten could stop scratching to chase a littermate across the room).

Development of Stereotypic

Movements in the Chick Embryo

The presence of an adultlike locomotor pattern in kittens within a day or two after birth suggests that the neural substrate for coordinating this pattern is established earlier in development. Because the chick embryo can be studied during normal behavior in its natural environment, it was selected to explore embryonic development of coordinated motor patterns. Electromyographic recordings at E9 and E10 indicated that there was already a distinct pattern of coordination among a group of leg muscles by E9 in normal embryos. [46] Specifically, flexors of the hip, knee, and ankle were synchronously activated and reciprocally activated with the extensors (Fig. 7). However, the pattern differed from that for locomotion in the post-hatching chick. Namely, burst durations for flexors and extensors were similar to one another, and neither burst was closely related to cycle period; in locomotion, the extensor burst is proportionately longer in duration than the flexor burst and is closely related to the cycle period, as observed in cats. [19,48] During episodes of spontaneous motility, there was visible variability in limb excursions and body movements. However, it appeared that the embryo's posture in ovo was continuously perturbed by both the spontaneous contractions of the allantoic membrane and the self-induced perturbations resulting from nonstabilized movement in an aqueous medium (amniotic fluid).

Effects of Spinal Transection on

the Development of Stereotypic

Movements

Hindlimb stepping and paw shaking appeared to be more readily initiated and sustained after spinal transection. In spinal kittens, airstepping was observed within 72 hours after spinal transection when the limbs were unloaded (pendent). [44] Similarly, hindlimb stepping was readily initiated when the limbs contacted the moving belt of the treadmill, but no weight support was accepted during initial postnatal recovery from the spinal transection. Weight support gradually improved over the first postnatal month, but few spinal kittens demonstrated full weight support in the hindlimbs during the first 2 postnatal months. Spinal kittens were never observed to use the hindlimbs to step overground. Initially, paw shaking was also more readily observed in spinal kittens and was evoked approximately 2 weeks in advance of normal littermates. To see the response, however, it was necessary to support the kitten with the hindlimbs pendent, as during airstepping. In spinal chick embryos, spontaneous motility also appeared to be more readily sustained following spinal transection in that spinal embryos performed more continuous cycles of kicking than normal embryos.

The corresponding EMG patterns, however, differed from normal patterns. Despite the apparent early emergence or sustained expression of repetitive behaviors in the hindlimbs of neonatal kittens [44,45] (Fig. 3B) and chick embryos [47] (NS Bradley and A Bekoff, manuscript in progress) following spinal transection, EMG patterns differed from those for normal animals. Furthermore, in the studies of kittens, EMG patterns of spinal kittens also differed from those of normal adult cats and cats spinal-transected as adults. For example, the EMG patterns for stepping in spinal kittens and for spontaneous motility in chick embryos were characterized by shorter cycle periods and burst durations than in normal animals of the same age. In spinal kittens, the relationship between antagonist muscles was also altered; extensor burst duration did not vary closely with cycle period. The pattern also differed from normal in that burst duration for flexors and extensors was similar (Fig. 2B) of the burst duration for flexors was greater than for extensors (Fig. 3B).

Disruptions were also noted in EMG records for paw shaking in the spinal kittens. [45] The mixed synergy (TA + VL) was observed in only one response out of a total of 92 analyzed. In all other instances, the VL was coactive with LG (Fig. 6B). There was also a notable difference in the temporal organization across consecutive cycles. In normal littermates, as in adults, there was a progressive increase in burst duration across consecutive cycles; in spinal kittens, there was no apparent temporal organization across cycles.

Discussion

Early Development of

Coordinated Motor Patterns

As evidenced in the chick embryo and indirectly suggested in stepping records for neonatal kittens, at least some coordinated patterns of muscle activity used in adult behaviors were established very early in embryonic development. Namely, synergists were coactivated and reciprocally activated with antagonists, and, in the embryo, the pattern was a distinct one of equal and alternating activity in the flexors and the extensors. Furthermore, it appears that the pattern was appropriately modified to produce the different patterns for locomotion, scratching, and paw shaking during a neonate's first performances of each behavior, despite some awkwardness as it attempted to maintain a postural base. Thus, for the hindlimb behaviors in the animals studied, the extended period of motor development did not appear to be characterized by a process of assembling coordinated motor patterns. Rather, the developmental period appeared to be characterized by a process of acquiring the postural skills to support and adapt each behavior.

The clinician may ask to what extent, if any, do these findings apply to human development? To date, studies of human neonates using surface electrodes to study infant stepping or kicking [51] or early locomotion [52] have not detected distinct EMG patterns. Yet, the literature contains several reports of limb movements in fetuses [33,34] and neonates, [53] including visually triggered reaching, [39] that suggest there is some potential to organize coordinated movement very early in human development. Failure to observe coordinated muscle patterns may be due to the hardware and EMG-processing procedures used. Additionally, failure to observe coordinated patterns may be due to testing conditions that are insufficient to evoke the patterns, conditions that mask the patterns by requiring secondary movements or postural adjustments, or conditions that inhibit expression of the patterns.

One lesson to be learned from the animal studies is that the testing context is a powerful variable that may or may not reveal an animal's potential abilities. For example, during initial postnatal performances, the ankle antagonists were synchronously active during the extensor portion of the cycle for weight-supported steps (Fig. 2A). The cycle periods for these steps were as much as 3 to 5 seconds in duration and considerably longer than cycle periods of 1 to 2 seconds during the second postnatal month. Whereas, during hindlimb stepping without weight support (airstepping), cycle periods were shorter and the adult pattern of reciprocal activation was more apparent (Fig. 3A). These findings suggest that the pattern of muscle activation varied with the postural demand of the task and that synchronous activation of the ankle antagonist muscles during sustained weight support may have been an adaptive strategy to provide stiffness about the ankle to compensate for fatigue and/or insufficient postural control by higher neural centers. The fact that contractile properties for ankle muscles are immature at birth and differentiation of fiber types is not complete until 6 to 8 weeks would appear to support this interpretation. [54] Furthermore, the fact that EMG records for stepping in age-matched spinal kittens exhibited reciprocal activity in the ankle antagonist muscles during hindlimb stepping characterized by cycle periods of 1 second and little or no weight support would also appear to support this interpretation.

Mechanical Factors Critical for

Expression of Coordinated

Motor Patterns

In Hamburger's [55] classic study of embryonic motility, it was concluded from visual observations that there was no coordination between body parts during these movements until the embryo began prehatching behavior in the final days of embryonic development. It is likely, however, that the discrepancy between the observations of movement and the early emergence of distinct EMG patterns is due to specific environmental factors. For example, during the early stages of development, the embryo is submerged in an aqueous environment (amniotic fluid) and no part of its body comes in contact with the shell. In this buoyant condition, forces are differently distributed during movement than the gravitational and frictional forces encountered when buoyancy is eliminated and the feet are in contact with a surface. Thus, when the buoyant embryo initiates a series of kicks, reaction forces are likely to contribute to postural displacements characterized by rocking of the embryo in a broad rostrocaudal arc about the vessels attaching it to the yolk sac. The postural displacements, in turn, are likely to mechanically perturb ongoing movements and alter the phasing between the different body segments. In summary, it appears that the developing nervous system can organize coordinated action by appropriately activating and phasing different motoneuron pools; however, the resulting movements appear to lack coordination, because mechanical forces perturb the relative phasing between body segments. An analogy perhaps more familiar to the reader is the awkward movements made by astronauts in a zero-gravity environment as they attempt to use the motor strategies they normally use to move against gravity on earth.

It is also likely that mechanical factors contribute to the differences observed in EMG records during postnatal development for stepping and paw shaking in normal and spinal kittens. Kinetic data for adult cats suggest that mechanical factors and resulting motion-dependent feedback play a significant role in determining the muscle patterns for both locomotion and paw shaking. [56] For example, previous studies [57,58] have shown that extensor burst duration is a function of weight support during locomotor behaviors; the longer the stance phase or tension maintained in the ankle extensor muscles, the longer the extensor burst duration. Conversely, in spinal kittens, the altered EMG patterns for stepping corresponded to treadmill locomotion characterized by little or no weight support. Similarly, during paw shaking in normal cats, sensory feedback associated with repeated cycles of ballistic flexion and extension at the ankle and knee has been shown to be related to the production of a novel muscle pattern called the mixed synergy (synchronous activation of the TA and VL muscles). [56] The absence of the mixed synergy in paw-shaking responses for spinal kittens was likely due to a failure to achieve the high velocities or to appropriately phase the peak velocities at the ankle and knee. One possible explanation for the failure to attain appropriate accelerations, as well as the reduced weight support observed in spinal kittens, is that the differentiation of motor unit fiber types is altered following transection during development. [59] The poorly differentiated units are characterized by reduced force-producing capabilities that likely limit the ability of muscle to generate large accelerations and to meet normal weight-support requirements.

Clinical Implications

Until we know more about the potential motor abilities of the human fetus and neonate, it is difficult to determine to what extent animal models and the findings such as those reported in this article can be translated into clinical practice. Thus, one implication is the need for clinicians to participate in extending our knowledge of normal development and to study the many factors, including anatomical, physiological, biomechanical, psychological, etc, that contribute to the processes that drive motor development. Every discipline of study has its biases that advance and hinder the birth of new insights. In the studies presented, a traditional neurological point of view does not explain discrepancies between neural activity and behavioral outcome or the processes that link the two. To further understand the processes, it will be important to learn more about the mechanics that act upon the developing body as it moves in an environment governed by physical laws. Because behavior begins before a musculoskeletal system is fully differentiated, it will also be necessary to learn more about the unique contributions this maturing system makes to motor control at different stages of development.

Another important implication for clinical practice is the need to develop useful and reliable quantitative measures of performance. In the studies presented, the use of a repetitive behavior expedited the measurement of one movement-related variable (EMG) and identification of developmental changes. By using video records, it was also possible to review the apparent behavior and to identify other factors important for understanding the EMG measurements obtained. In the future, it will be important to obtain specific measurements of body movements synchronized with EMG, not only to know the details of a behavior as it is refined during development, but also to understand the dynamics (forces) that shape the movement, so as to gain an understanding of the processes that establish motor control. Perhaps one of the most important clinical implications to be drawn from motor control studies is that a biomechanical assessment of movement dynamics can greatly advance our understanding of specific movement problems (for further discussion, see Smith and Zernicke [56]. In some instances, contextual forces may mask the potential to produce movement. In other instances, the forces may act to create a movement different from that intended, or they may require the nervous system to adapt the motor pattern initially executed. For example, when cats locomote, the sartorius muscle produces a two-burst pattern, one burst during late stance and a second burst at the end of swing. [56]. The latter burst, however, is not always present, especially at slower walking speeds. An assessment of the kinetics and kinematics suggests that at high walking speeds, the forces that accelerate the knee into extension are very large and must be countered by flexor muscle activity. At slower walking speeds, however, it appears that passive elastic forces of the knee flexors and connective tissues are sufficient to counteract acceleration of the knee into extension. [56] Thus, opposing flexor forces are not required, and the motor pattern does not include a second burst of sartorius muscle activity. Therefore, depending on the dynamics of the movement, the motor pattern will vary; it will retain the basic features of the motor pattern that typifies walking, but will also contain modifications to meet the contextual (ie, force) requirements.

Conclusion

The studies of motor development in neonatal kittens and chick embryos demonstrate that at least some motor patterns for repetitive behaviors are established very clearly in development and can appear adultlike even during the very first performance of the behavior. Findings in normal and spinal-transected animals suggest, however, that many different variables may determine whether coordinated behaviors are produced. Most notable among these are context-dependent variables, such as postural requirements, that are required to initiate the behavior and mechanical variables that shape movement and determine motion-dependent feedback. In the future, it will be important to more precisely identify those variables that contribute to expression or masking of potential motor abilities to extend our knowledge of the processes that determine development and to effectively treat movement problems that arise during development.

NS Bradley, PhD, PT, is Assistant Professor, School of Physical and Occupationsl Therapy, McGill University, 3654 Drummond St, Montreal, Quebec, Canada H3G 1Y5.

The work is a summary of previously published studies supported by National Institutes of Health grants NS 19684 to JL Smith, NS 20310 to A Bekoff, and T32 HDO7288 (postdoctoral traineeship) to NS Bradley.

References

[1] Bradley NS, Bekoff A. Development of locomotion: animal models. In: Woollacott M, Shumway-Cook A. eds. The Development of Posture and Gait Across the Lifespan. Columbia, SC: University of South Carolina Press; 1989:48-73.

[2] Prechtl HFR, Beintema D. The Neurological Examination of the Full-Term Newborn. Philadelphia, Pa: JB Lippincott Co; 1975.

[3] Dubowitz V, Dubowitz L. The Neurological Assessment of the Preterm and Full-Term Newborn Infant. Philadelphia, Pa: JB Lippincott Co; 1981.

[4] Brazelton TB. Neonatal Behavioral Assessment Scale. Philadelphia, Pa: JB Lippincott Co; 1973.

[5] Sherrington C. The Integrative Action of the Nervous System. New York, NY: Macmillan Publishing Co; 1961.

[6] Grillner S. Control of locomotion in bipeds, tetrapods, and fish. In: Geiger SR, ed: Handbook of Physiology. Bethesda, Md: American Physiological Society; 1981:1179-1236.

[7] Langworthy OR. A correlated study of the development of reflex activity in fetal and young kittens and the myelinization of tracts in the nervous system. Contrib Embryol Carneg Inst. 1929;20:127-171.

[8] Windle WF, Griffin AM. Observations on embryonic and fetal movements of the cat. J Comp Neurol. 1931;52:149-188.

[9] Humphrey T. Some correlations between the appearance of human reflexes and the development of the nervous system. Prog Brain Res. 1964; 4: 93-135.

[10] Hooker D. Evidence of Prenatal Function of the Central Nervous System in Man. New York, NY: American Museum of Natural History; 1958.

[11] Saint-Anne Dargassies S. Normal and pathological fetal behavior as seen through neurological study of the premature newborn. Contrib Gynecol Obstet. 1979; 6: 42-56.

[12] Allen MC, Capute AJ. The evolution of primitive reflexes in extremely premature infants. Pediatr Res. 1986; 20: 1284-1289.

[13] Allen MC, Capute AJ. Neonatal neurodevelopmental examination as a predictor of neuromotor outcome in premature infants. Pediatrics. 1989; 83: 498-506.

[14] Capute AJ, Palmer FB, Shapiro BK, et al. Primitive reflex profile: a quantitation of primitive reflexes in infancy. Dev Med Child Neurol. 1987; 26: 375-383.

[15] Touwen BCL. Primitive reflexes: conceptual or semantic problem? In: Prechtl HFR, ed. Continuity of Neural Functions from Prenatal to Postnatal Life. Philadelphia, Pa: JB Lippincott Co; 1984: 115-125.

[16] Vilensky JA, Gankiewicz E, Gehlsen G. A kinematic comparison of backward and forward walking in humans. Jornal of Human Movement Studies. 1987; 13: 29-50.

[17] Boylls CC, Zomlefer MR, Zajac FE. Kinematic and EMG reactions to imposed interlimb phase alterations during bipedal cycling. Brain Res. 1984; 324: 342-345.

[18] Deliagina TG, Feldman AG, Gelfand IM, Orlovsky G. On the role of central program and afferent inflow in the control of scratching movements in the cat. Brain Res. 1975; 100: 297-313.

[19] Halbertsma JM: The stride cycle of the cat: the modeling of locomotion by computerized analysis of automatic recordings. Acta Physiol Scand [Suppl]. 1983; 521.

[20] Buford J, Smith JL, Zernicke RF. Hindlimb muscle synergies during backward walking in the cat. Society for Neuroscience Abstracts. 1989; 15: 392. Abstract.

[21] Carlson Kuhta P, Smith JL, Zernicke RF. Hindlimb kinematics and muscle synergies during scratching. Society for Neuroscience Abstracts. 1989; 15: 394. Abstract.

[22] Gruner JA, Altman J. Swimming in the rat: analysis of locomotor performance in comparison to stepping. Exp Brain Res. 1980; 40: 374-382.

[23] Stehouwer DJ, Farel PB. Development of hindlimb locomotor behavior in the frog. Dev Psychobiol. 1984; 17: 217-232.

[2'] Mortin LI, Keifer J, Stein PSG. Three forms of the scratch reflex in the spinal turtle: movement analyses. J Neurophysiol. 1985; 53: 1501-1516.

[25] Bekoff A, Nusbaum MP, Sabichi AL, Clifford M. Neural control of limb coordination: I. Comparison of hatching and walking motor output patterns in normal and deafferented chicks. J Neurosci. 1987; 7: 2320-2330.

[26] Jacobson RD, Hollyday M. A behavioral and electromyographic study of walking in the chick. J Neurophysiol. 1982; 48: 238-256.

[27] Berkinblit MB, Deliagina TG, Orlovsky GN, Feldman AG. Activity of motoneurons during fictitious scratch reflex in the cat. Brain Res. 1980; 193: 427-438.

[28] Robertson GA, Mortin LI, Keifer J, Stein PSG. Three forms of the scratch reflex in the spinal turtle: central generation of motor patterns. J Neurophysiol. 1985; 53: 1517-1534.

[29] Grillner S, Zangger P. On the central generation of locomotion in the low spinal cat. Exp Brain Res. 1979; 34: 241-261.

[30] Stehouwer DJ, Farel PB. Development of locomotor mechanisms in the frog. J Neurophysiol. 1985; 53: 1453-1466.

[31] Bekoff A. Ontogeny of leg motor output in the chick embryo: a neural analysis. Brain Res. 1976; 106: 271-291.

[32] Watson A, Bekoff A. A kinematic analysis of hindlimb motility in 9- and 10-day old chick embryos. Society for Neuroscience Abstracts. 1987; 13: 1504. Abstract.

[33] Ianniruberto A, Tajani E. Ultrasonographic study of fetal movements. Semin Perinatol. 1981; 5: 175-181.

[34] de Vries JIP, Visser GHA, Prechtl HFR. The emergence of fetal behavior: I. Qualitative aspects. Early Hum Dev. 1982; 7: 301-322.

[35] Provine RR. Development of between-limb movement synchronization in the chick embryo. Dev Psychobiol. 1980; 13: 151-163.

[36] Bekoff A, Trainer W. The development of interlimb co-ordination during swimming in postnatal rats. J Exp Zool. 1979; 83: 1-11, 1979.

[37] Fentress JC. Order in ontogeny: relational dynamics. In: Immelmann K, Barlow GW, Petrinovitch L, et al, eds. Behavioral Development: The Bielefeld Interdisciplinary Project. Cambridge, England: Cambridge University Press; 1981: 338-371.

[38] Bayley N. The Bayley Scales of Infant Development. New York, NY: The Psychological Corp; 1969.

[39] von Hofsten C. Eye-hand coordination in the newborn. Dev Psychol. 1982; 18: 450-461.

[40] Bekoff A, Kauer JA. Neural control of hatching: fate of the pattern generator for the leg movements of hatching in posthatching chicks. J Neurosci. 1984; 4: 2659-2666.

[41] Thelen E. Learning to walk in still an "old" problem: a reply to Zelaro (1983). Journal of Motor Behavior. 1983; 15: 139-161.

[42] Zelaro PR. The development of walking: new findings and old assumptions. Journal of Motor Behavior. 1983; 15: 99-137.

[43] Bradley NS, Smith JL. Neuromuscular patterns of stereotypic hindlimb behaviors in the first two postnatal months: I. Stepping in normal kittens. Dev Brain Res. 1988; 38: 37-52.

[44] Bradley NS, Smith JL. Neuromuscular patterns of stereotypic hindlimb behaviors in the first two postnatal months: II. Stepping in spinal kittens. Dev Brain Res. 1988; 38: 53-67.

[45] Bradley NS, Smith JL. Neuromuscular patterns of stereotypic hindlimb behabiors in the first two postnatal months: III. Scratching and the paw-shake response in kittens. Dev Brain Res. 1988; 38: 69-82.

[46] Bradley NS, Bekoff A. Development of coordinated movement in chicks: I. Temporal analysis of hindlimb muscle synergies at embryonic days 9 and 10. Dev Psychobiol. In press.

[47] Bradley NS, Bekoff A. Hindlimb synergies in spinal chick embryos. Society for Neuroscience Abstracts. 1988; 14: 1003. Abstract.

[48] Forssberg H, Grillner S, Halbertsma J. The locomotion of the low spinal cat: I. Coordination with a hindlimb. Acta Physiol Scand. 1980; 108: 269-281.

[49] Smith JL, Hoy MG, Koshland GF, et al. Intralimb coordination of the paw-shake response: a novel mixed synergy. J Neurophysiol. 1985; 54: 1271-1281.

[50] Hamburger V, Balaban M, Oppenheim R, Wenger E. Periodic motility of normal and spinal chick embryos between 8 and 17 days of incubation. J Exp Zool. 1965; 159: 1-14.

[51] Thelen E, Fisher DM. Newborn stepping: an explanation for a "disappearing" reflex. Dev Psychol. 1982; 18: 760-775.

[52] Forssberg H. A developmental model of human locomotion. In: Grillner S, Stein PSG, Stuart DG, et al, eds. Neurobiology of Vertebrate Locomotion. London, england: Macmillan Publishers LTD; 1986; 485-501.

[53] Thelen E, Ridley-Johnson R, Fisher DM. Shifting patterns of bilateral coordination and lateral dominance in the leg movements of young infants. Dev Psychobiol. 1983; 16: 29-46.

[54] Hammarberg C, Kellerth J-O. The postnatal development of some twitch and fatigue properties of the ankle flexor and extensor muscles of the cat. Acta Physiol Scand. 1975; 95: 166-178.

[55] Hamburger V. Some aspects of the embryology of behavior. Q Rev Biol. 1963; 38: 342-365.

[56] Smith JL, Zernicke RF. Predictions for neural control based on limb dynamics. Trends in Neuroscience. 1987; 10: 123-128.

[57] Duysens J, Pearson KG. Inhibition of flexor burst generation by loading ankle extensor muscles in walking cats. Brain Res. 1980; 187: 321-332.

[58] Giuliani CA, Smith JL. Development and characteristics of airstepping in chronic spinal cats. J Neurosci. 1985; 5: 1276-1282.

[59] Gallego R, Huizar P, Kudo N, Kuno M. Disparity of motoneurone and muscle differentiation followinf spinal transection in the kitten. J Physiol. 1978; 281: 253-265.