Interlimb coordination in human crawling reveals similarities in development and neural control with quadrupeds

Abstract

The study of quadrupeds has furnished most of our understanding of mammalian locomotion. To allow a more direct comparison of coordination between the four limbs in humans and quadrupeds, we studied crawling in the human, a behavior that is part of normal human development and mechanically more similar to quadrupedal locomotion than is bipedal walking. Interlimb coordination during hands-and-knees crawling is compared between humans and quadrupeds and between human infants and adults. Mechanical factors were manipulated during crawling to understand the relative contributions of mechanics and neural control. Twenty-six infants and seven adults were studied. Video, force plate, and electrogoniometer data were collected. Belt speed of the treadmill, width of base, and limb length were manipulated in adults. Influences of unweighting and limb length were explored in infants. Infants tended to move diagonal limbs together (trot-like). Adults additionally moved ipsilateral limbs together (pace-like). At lower speeds, movements of the four limbs were more equally spaced in time, with no clear pairing of limbs. At higher speeds, running symmetrical gaits were never observed, although one adult galloped. Widening stance prevented adults from using the pace-like gait, whereas lengthening the hind limbs (hands-and-feet crawling) largely prevented the trot-like gait. Limb length and unweighting had no effect on coordination in infants. We conclude that human crawling shares features both with other primates and with nonprimate quadrupeds, suggesting similar underlying mechanisms. The greater restriction in coordination patterns used by infants suggests their nervous system has less flexibility.

FIG. 1.

Examples of coordination illustrated with limb contact patterns. Sequences of limb contact from 3 different subjects: one infant (trot-like) and 2 adults (no limb pairing, and pace-like) crawling on the treadmill. Time is represented horizontally, with periods of stance (solid lines) and swing (spaces). In the trot-like gait, the diagonal limbs enter stance close together in time. In the pace-like gait, the ipsilateral limbs enter stance close together in time. There is said to be no limb pairing when the initiations of stance in all the limbs are equally spaced in time. Ipsilateral phase lag, the primary measure of coordination, is the delay between left arm contact and the preceding left leg contact (b), expressed as a percentage of the cycle duration of the left leg (a).

FIG. 2.

Coordination of limbs during symmetrical hands-and-knees crawling. Histograms describe the incidence of initiation of stance and swing (vertical axis) for each of the 4 limbs with respect to the step cycle of the left leg (horizontal axis). Data are from infants crawling overground (A, 94 cycles from 13 infants) and adults crawling on the treadmill (B, C, and D). Two adults showed a trot-like coordination (B, 194 cycles), one showed a pace-like coordination (C, 177 cycles), and one showed all forms of coordination (D, 225 cycles). Initiation of stance of the left knee occurs at 0 and 100% of the step cycle (vertical dotted lines). Gray bars: swing; black bars: stance. Bin width: 2%.

FIG. 3.

Effect of rate of crawling on ipsilateral phase lag. Data are individual cycles from infants crawling overground (A, 94 cycles) and adults crawling on the treadmill (B and C). Adult data are divided into those who showed a specific preference for one form of coordination (B, trot-like coordination: 2 subjects, filled circles, 224 cycles; pace-like coordination: one subject, open circles, 188 cycles), and the one subject who did not (C, 242 cycles). Infants showed a weak, linear relationship between rate of crawling and pattern used (r² = 0.17, P < 0.05). Adults showed no simple relationship with crawling rate.

FIG. 4.

Coordination in human crawling compared with quadrupeds. The ipsilateral phase lag is plotted against duty factor of the leg for phase lags <25% or >50%, and against duty factor of the arm for phase lags between 25 and 50%, in accordance with the model presented by Cartmill et al. (2002). Prediction lines from the Cartmill model are superimposed on our data. The top prediction line describes typical nonhuman primate behavior. The bottom 2 prediction lines fit data from nonprimate quadrupeds. Infant data (open circles, 39 cycles) are scattered around the prediction line for nonprimates that prefer trot-like coordinations. Adult data (filled circles, 469 cycles) fit the shape of the 2 prediction lines for nonprimates, but is shifted from the model.

FIG. 5.

Mechanical influences on patterns of coordination. A: unobstructed crawling of 5 adults at 0.45 m/s exhibited a wide range of coordination patterns (filled circles, 99 cycles). Slowing the treadmill to 0.13 m/s restricted crawling to coordinations with little or no pairing of limbs (open circles, 100 cycles). B: an obstruction along the midline limited adults to largely trot-like coordination at the higher speed (filled circles, 50 cycles) and had little effect on coordination pattern at the lower speed (open circles, 50 cycles) compared with A. C: unweighting of infants by 43 to 88% (filled triangles, 73 cycles) did not change coordination patterns used compared with when infants supported most (66–87%) of their own weight (open triangles, 100 cycles). D: the midline obstruction imposed on the adults forced a significantly wider stance at both the arms and the legs (indicated by an asterisk [*]); resultant stance widths were similar to those of infants during undisturbed overground crawling (ANOVA with Tukey post hoc test). Bars are mean ± SD across subjects. Stance width was normalized to shoulder width.

FIG. 6.

Ipsilateral phase lag in hands-and-feet crawling. Two infants crawling on hands and feet (open triangles, 29 cycles) demonstrated the same coordination patterns as those of infants crawling on hands and knees (i.e., Fig. 3A). Five adults crawling on hands and feet showed predominantly no clear pairing of limbs, or pairing of ipsilateral limbs (filled triangles, 161 cycles).

FIG. 7.

Crawling sequences with transitions in phasing between limbs. A: limb contact patterns showing transition from pace-like cycles into a rotary gallop. The arrow indicates the beginning of the transition. The graphs in B demonstrate the changes in rate of crawling and phasing between limb pairs (legs, arms, ipsilateral) that occurred during the 2 sequences in which the subject transitioned to a gallop. Filled circles represent the sequence shown in A; open squares represent a second sequence (not shown), in which the subject transitioned to a transverse gallop. C and D: transitions within symmetrical crawling sequences. Ipsilateral phase lag (vertical axis) for individual cycles was plotted for sample sequences during which there was a change in ipsilateral phasing by ≥10%. Plots present data from 2 infants crawling overground (C), and two adults on the treadmill (D).