Muscle Activation and Coordinated Movements of Infant Rolling

Abstract

Rolling is a critical step of infant development, encouraging muscle coordination and enabling independent exploration. Understanding muscle activity during infant rolling movements on a flat surface is necessary to more fully characterize how the rolling milestone is achieved. The purpose of this study was to determine infants’ muscle activation throughout roll initiation for six previously established coordinated movements. Thirty-eight healthy infants (age: 6.5±0.7 months; 23M/15F) were enrolled in this IRB-approved in-vivo biomechanics study. Surface electromyography sensors recorded muscle utilization from the erector spinae, abdominal muscles, quadriceps, and hamstrings while infants rolled. Each rolling movement was categorized as one of six roll types, and the mean muscle activity was analyzed. All roll types required initial activation of all measured muscle groups. Movements featuring axial rotation of the torso relative to the pelvis required highly active erector spinae muscles. Movements featuring trunk and hip flexion required highly active abdominal muscles. Infants used distinct coordinated muscle activations to achieve the six different roll types on a flat surface. A foundational understanding of the different muscle activation patterns required during infant rolling will provide crucial insight into motor development. This study quantified muscle coordination required of infants to achieve rolling on a firm flat surface. Previous research indicates that the mechanical environment in which an infant is placed impacts muscle activity and body position during normal lying. Therefore, future work should explore if mechanical environments that differ from a flat and firm surface also influence these coordinated movements and muscle activations.

1. INTRODUCTION

Achieving a roll is a key motor skill and an important developmental milestone for infants. To initiate and complete a roll, infants must use whole-body, goal-oriented movements that take them from a supine to prone or prone to supine position (Richter et al., 1989). Rolling also requires an initial disruption of an otherwise stable body by using coordinated segmental body movements. During these movements, limbs interact with the rolling surface to provide momentum and move freely to alter stability and balance (Altai et al., 2019; Kobayashi et al., 2016, 2021). Infants begin rolling from supine to prone between four to six months as they develop their gross motor control (Altmann and Hill, 2019; Colvin et al., 2014), but they begin learning the skills needed to roll as soon as they are born (McGraw, 1941). Rolling encourages the muscle development and postural control needed to learn and achieve more advanced movements like sitting, crawling, and walking (Altmann and Hill, 2019). Therefore, it is important to determine how infants use and develop their muscles during rolling to ensure proper motor development and improve commercial infant products.

Motor abilities, like rolling, are part of the cognitive and motor development of infants through interaction with others and by exploring their environment (Lobo et al., 2013). Assessment of motor abilities and infant movement patterns are an important component in the early identification of motor delays (Harris and Heriza, 1987). For example, one of the early signs of hypertonia and cerebral palsy is when an infant can achieve a roll early, but the maneuver is sudden and stiff with little control (Bennett, 1999; Burns et al., 1989). Early intervention of possible motor delays can have a significant impact on a child’s ability to learn new skills and increase their success in school and life (CDC, 2022). It is estimated that approximately 15% of children have developmental disabilities that would benefit from physical therapy interventions (Boyle et al., 2011), however, only 3.1% of infants and toddlers receive the early interventions they need (ECTA, 2022). Understanding the muscle activation and coordinated movements of rolling in healthy infants could assist clinicians in the early identification of motor delays and guide physical therapy on muscle development for infants who are not yet rolling.

In addition to aiding guided interventions, characterizing rolling development can inform the juvenile products industry by providing a greater understanding of infant movements and muscle utilization. Our previous study, looking solely at inclined crib mattress environments, determined that the muscle utilization to maintain a prone posture on an incline was significantly different compared to the flat surface (Mannen et al., 2019; Wang et al., 2020). This indicates that other movements, like rolling, may also exhibit changes in muscle utilization in different mechanical environments. However, a baseline of infant rolling on a flat surface must first be established.

Previous research has determined that infants primarily use six unique coordinated movements to achieve a supine to prone roll (Kobayashi et al., 2016). Each of the six different coordinated movements are described in Figure 1 where the limbs are codified as ipsilateral arm (IA) and leg (IL) and contralateral arm (CA) and leg (CL) in the direction of the roll. Further research on the kinetics of these coordinated movements has demonstrated that an infant’s loading patterns of each limb differ between movements, even if the motion is similar. This indicates that certain movements require more support from the ground during rolling. For example, coordinated movements B, D, and E use a strategy where the following limb pushes on the ground to achieve a roll, while coordinated movements A, C, and D do not load the stationary limbs indicating they may be utilized more for balance than support (Kobayashi et al., 2021).

FIGURE 1:

Six previously established coordinated movements of rolling as defined by Kobayashi et. al. (2016), figures and captions adapted and reproduced with permission. Illustrated rolls are to the infants left. The limbs marked with an asterisk (*) are stationary throughout the roll, the dark gray limbs initiate the roll, and the light gray limbs follow after the roll has already been initiated.

Although these coordinated movements have been defined through kinematics and kinetics, the muscle activation patterns are still unknown. Characterizing muscle coordination used to achieve rolling may elucidate key motor development trajectories which could be used to assist clinicians in early identification of motor delays and guide therapy when infants are not meeting milestones. Furthermore, some design features of commercial infant products may facilitate or hinder infant rolling (Wang et al., 2020, 2021). A more thorough understanding of how an infant achieves a roll on a flat surface will provide crucial insight into what product design features may influence rolling. The purpose of this study was to determine the muscle activations throughout infant roll initiation for the six established coordinated movements as defined by Kobayashi et al. (2016). We hypothesized that distinct muscle activation characteristics would be identified for each of the six different movements used to achieve a roll.

2. METHODS

2.1. Participants

We recruited legal guardians of potential participants through flyers and social media after institutional review board approval. Infants between the ages of four and seven months, >37 weeks gestation, between the 5^th and 95^th percentile for birth height and weight, and who had successfully achieved at least two independent supine to prone rolls were included in this study. Infants were excluded if they had any diagnosed orthopaedic or neurological conditions that may cause motor or developmental delays. Before participating, guardians provided informed assent and completed an Ages and Stages Questionnaire (ASQ), a screening tool that monitors development (Squires et al., 2009). Infants were excluded from the study if they scored below the gross motor category cutoff for their age range on the ASQ (6-month cutoff score = 22.25) where further assessment by a professional is recommended.

2.2. Experimental Procedure

To determine the muscle activation of the six previously defined coordinated movements (Kobayashi et al., 2016), infants were placed in a supine position on a firm foam playmat for five-minutes or until they achieved a complete supine to prone roll (Figure 2). Infants were encouraged to roll in either direction (to their right or to their left) by their guardians or researchers offering toys, food, or simply calling out to them. As was done by Kobayashi et al. (2016), we analyzed the supine to lateral rotation portion of the roll, defined hereafter as a rolling movement. If infants did not achieve at least one rolling movement during the five-minute trial, they were excluded from further analysis. Any rolling movements assisted by parents or researchers, or where an infant did not begin in a flat supine position, were also excluded.

FIGURE 2:

Experimental setup is shown on the left and EMG placement for the abdominal muscles (AB), quadriceps (QUAD), erector spinae (ES), and hamstrings (HAM) is shown on the right. The ES sensors were placed mid back on either side of the spine. The QUAD sensors were placed midway between the knee and the anterior superior iliac spine aligned with the femur bone. The HAM sensors were placed directly behind the QUAD sensors on the posterior side of the leg. Finally, the AB sensors were placed approximately two finger widths apart just above the belly button. EMG was also collected from the cervical paraspinals, gluteus maximus, and triceps brachii muscles. However, these muscle groups were not discussed as significant contributions were not found both in this study and through previous research (Kobayashi et al., 2021; Wang et al., 2020). A kinematic data set was also collected but was outside of the scope of analysis for this study.

2.2.1. Video

Each trial was recorded (GoPro, San Mateo, CA) to visually determine the coordinated movement used. The videos were reviewed after testing and truncated to only include each rolling movement. Using our previously verified methodology for identifying the six coordinated movements through video alone (Siegel et al., 2023), a single trained reader determined the coordinated movement of each video. A one-factor analysis of variance (ANOVA) was used with six levels for each of the coordinated movements to determine if there was a statistical difference between the coordinated movement type and age as well as coordinated movement and total time to achieve the roll (SPSS Inc., version 26; Chicago, IL). If significance was found, a Dunnett’s T3 post-hoc test (p<0.05) was performed.

2.2.2. Electromyography

Surface EMG electrodes (Delsys, Natick, MA; 2000 Hz) recorded bilateral muscle activity from the erector spinae (ES), hamstrings (HAM), abdominal muscles (AB), and quadriceps (QUAD) (Figure 2). Each sensor was placed by the same researchers and wrapped in self-adherent bandage to ensure placement and consistency between participants.

The EMG was extracted and truncated to only include each rolling movement. In addition, a five-second resting period, where the infant was lying still in a supine position, was extracted for each infant to use for normalization. Maximal voluntary isometric contractions, which are usually used to normalize EMG in children or adults, are impossible to obtain in an infant population. A time period of five-seconds was chosen based on previous adult EMG studies (Gonzalez et al., 2015; Runciman et al., 2015). Using MATLAB, raw EMG waveforms were assessed using a power spectral analysis to exclude corrupted or missing data. To reduce contamination from movement artifacts, electrocardiogram signals, and high frequency noise, the raw EMG waveforms were filtered using a band-pass 4^th order Butterworth filter between 35–500 Hz. The EMG waveforms were also notch-filtered at 60 Hz using a 4^th order Butterworth filter to eliminate interference from nearby electronic sources. The EMG waveforms were then full-wave rectified, baselined, and subjected to a low-pass 4^th order Butterworth filter (cutoff 50 Hz) to obtain the EMG envelope (Siddicky et al., 2020, 2021; Wang et al., 2020, 2021). This filtering process was completed for each rolling movement and five-second resting period. We used the interquartile range (IQR) method to remove sporadic errors (Thamsuwan and Johnson, 2022) and the muscle activation levels during the rolling movement were divided by the mean EMG of a resting period to compare between subjects. Each rolling movement was then represented as a percentage of the entire rolling movement (0 to 100%) and categorized by muscle group and coordinated movement used. The mean EMG value for each muscle group for all infants was calculated at each time point throughout the rolling movement for each roll type. The mean EMG magnitudes were then normalized to the maximum mean value across all roll types for each muscle group to compare between coordinated movements. A comparison was made for each muscle group between coordinated movements. The analysis was repeated to compare the ipsilateral and contralateral muscle activation.

A > 50% threshold value represents periods of higher muscle activity (Howard et al., 2017), so the mean muscle activity within each section of the roll determined the muscle utilization level to be low (0–49%) moderate (50–74%) or high (>75%). Each rolling movement was divided into three sections: the beginning (0–24% of the rolling movement complete), middle (25–74%) and end (75–100%), to determine unique characteristics throughout each coordinated movement. The whole rolling movement (0–100%) was also considered.

2.2.3. Oxygen Saturation

Infants’ oxygen saturation (SpO₂) was monitored to ensure safety using a medical grade pulse oximeter attached to the infant’s right foot (Figure 2). Based on the clinical safety standard for SpO₂, if the infant’s SpO₂ reading was <95% for more than five seconds (Hunt et al., 1999; Vold et al., 2012), the testing was ended for that trial.

2.3. Statistical Analysis

Statistical analyses were performed using the SPSS statistical package (SPSS Inc., version 26; Chicago, IL). The differences between coordinated movements on mean muscle activation were tested using a repeated measures ANOVA. A one-factor ANOVA design was used with six levels representing the six coordinated movements to be compared. This ANOVA was conducted for each muscle group (ES, HAM, AB, and QUAD) and repeated over the four different time periods (beginning, middle, end, and whole). If significant effects were found, a Dunnett’s T3 post-hoc test (p<0.05) was performed to determine which coordinated movements had significantly different muscle activation levels during different time periods. Additionally, a repeated measures ANOVA with a Tukey HSD post hoc-test (p<0.05) was completed to compare the ipsilateral and contralateral muscle activation levels of each coordinated movement for each muscle group and time period to determine if certain coordinated movements displayed significant differences in contralateral versus ipsilateral muscle activation.

3. RESULTS

Thirty-eight healthy infants (age: 6.5±0.7 months; 23M/15F) were enrolled in this study; no infants were excluded due to their ASQ results. Twenty-four infants (age: 6.7±0.7 months; 14M/10F) achieved at least one rolling movement during data collection, resulting in seventy-two rolling movements that were used for further analysis. Table 1 shows the demographic and developmental screening results of our included participants.

TABLE 1:

Demographics of the twenty-four infants included in data analysis

	Mean	STD	Minimum	Maximum
Age (Months)	6.7	0.7	4.6	7.6
Gestational Age at Birth (Weeks)	39.1	1.2	37.0	41.7
Height (cm)	64.7	3.9	56.0	72.5
Mass (kg)	7.8	1.1	6.2	10.4
Ages & Stages Gross Motor	42.5	12.2	25.0	60.0
	Male	Female
Sex	14	10
	White	Asian	Hispanic
Ethnicity	19	4	1

3.1. Coordinated Movements

All six previously defined coordinated movements (Figure 1) were observed. Coordinated movements C and F, were observed the most often each resulting in 26.4% of total rolling movements completed. Coordinated movement A was observed the least often, accounting for only 6.9% of total rolling movements completed (Figure 3). No statistical significance was found between the different movements and age. On average, each rolling movement took 3.6±2.8 seconds to achieve, with no significant difference between the coordinated movement types.

FIGURE 3:

Percentage of total rolling movements completed for each coordinated movement with included mean age

3.2. Muscle Activation

To demonstrate the roll identification process for the EMG data, we evaluated one trial where an infant was visually at rest before initiating the roll. Following the EMG analysis detailed in Section 2.2.2, we plotted the resting time period along with the subsequent rolling movement to ensure that the muscle activation changed between the two activities (Figure 4). For the resting period, the muscle activation is low and is consistent for all muscle groups. The rolling movement shows changes in muscle activation as the movement progresses.

FIGURE 4:

Top graph represents raw EMG data for the left quadricep while the bottom graph shows the filtered EMG data used for analysis for all muscle groups. Each graph compares a resting period and a single rolling movement (coordinated movement F) of a single trial.

For coordinated movement A (Figure 5a), all muscle groups work together at approximately the same level during the first half of the movement. For the second half, the QUAD muscle activation continues to increase while the ES, AB, and HAM muscle groups decrease. Coordinated movement B (Figure 5b) has high ES muscle activation throughout the rolling movement, while the QUAD and HAM muscles continuously increase, and the AB muscles stay consistent. During coordinated movement C (Figure 5c), the AB muscles have the highest activation, followed by the QUAD, ES, and then HAM muscles. For coordinated movement D (Figure 5d), all muscle groups follow a similar trend throughout the rolling movement. Coordinated movement E (Figure 5e) exhibits high ES muscle activation throughout most of the rolling movement. The AB and HAM muscles are lower throughout with the HAM muscles having a spike in muscle activation towards the end. Finally, during coordinated movement F (Figure 5f) the QUAD, and AB muscles follow the same trend of muscle activation throughout the rolling movement, while the HAM and ES act at a lower level.

FIGURE 5:

Mean muscle activation throughout the rolling movement for each of the six different coordinated movements (A-F). Each muscle group is represented as a percentage of the maximum mean muscle activation. Regions of green represent a low muscle activation, yellow is moderate, and red is a region of high muscle activation. Infant drawings adapted from Kobayashi, et al. (2016) and reproduced with permission.

Most coordinated movements begin with all muscles in the low activation region for the beginning of the rolling movement, with the exception of movement C and D where the AB and ES are in the moderate region respectively (Table 2). During the middle section of the rolling movement, all coordinated movements have muscle activations moving into the moderate to high regions. Coordinated movements A and D are the only movements where no muscle groups remain in the low region during the middle portion of the roll and coordinated movements B, C, and F feature muscle groups moving into the high activation region (Table 2). At the end of the rolling movement, most coordinated movements maintain or increase their muscle activation. However, coordinated movement A features a decrease in muscle activation in the ES, HAM, and AB muscles (Table 2). Looking throughout the rolling movements, the AB muscle activation remained at a low level for both coordinated movements B and E and the HAM muscle activation remained in the low region for coordinated movement C. For coordinated movement D, the ES remained in the moderate region throughout the entire movement. No coordinated movements exhibited muscle activation in the high region throughout the entire rolling movement, however, on average across the rolling movement, coordinated movement C had AB muscle activation in the high region.

TABLE 2:

Level of mean muscle activation for the beginning (0–24%), middle (25–74%), end (75–100%), and whole (0–100%) sections of the rolling movement where green represents low, yellow represents moderate, and red represents high muscle activation.

	Beginning	Middle	End	Whole
A	ES	ES	ES	ES
	HAM	HAM	HAM	HAM
	AB	AB	AB	AB
	QUAD	QUAD	QUAD	QUAD
B	ES	ES	ES	ES
	HAM	HAM	HAM	HAM
	AB	AB	AB	AB
	QUAD	QUAD	QUAD	QUAD
C	ES	ES	ES	ES
	HAM	HAM	HAM	HAM
	AB	AB	AB	AB
	QUAD	QUAD	QUAD	QUAD
D	ES	ES	ES	ES
	HAM	HAM	HAM	HAM
	AB	AB	AB	AB
	QUAD	QUAD	QUAD	QUAD
E	ES	ES	ES	ES
	HAM	HAM	HAM	HAM
	AB	AB	AB	AB
	QUAD	QUAD	QUAD	QUAD
F	ES	ES	ES	ES
	HAM	HAM	HAM	HAM
	AB	AB	AB	AB
	QUAD	QUAD	QUAD	QUAD

When dividing the limbs into contralateral and ipsilateral segments, few statistically significant differences for the varying coordinated movements were found in part due to the reduced sample size. After completing an ANOVA with a Tukey HSD post-hoc test, only the ES muscle activation for coordinated movement B (p=0.002) and the AB muscle activation for coordinated movement C (p=0.031) showed significant differences between the two sides of the body. Additionally, no meaningful trends were observed in the data analysis when comparing similar limb patterns between coordinated movements.

3.3. Statistical Analysis

The AB muscles had the most significant differences when comparing coordinated movements (Table 3). For the middle of the movement, the AB muscle activation was significantly higher for coordinated movement compared to both B (p<0.005) and E (p<0.005), coordinated movement E compared to D (p=0.03), and coordinated movement F compared to both B (p=0.04) and E (p<0.005). During the end section, movement E was significantly lower compared to movements C (p<0.005), D (p=0.02), and F (p<0.005). Over the whole rolling movement, movement C had significantly higher AB muscle activation levels compared to movements B (p<0.005), D (p=0.04), and E (p<0.005). Coordinated movement E had significantly lower AB muscle activation levels compared to movements D (p=0.01) and F (p<0.005).

Table 3:

Statistical outcomes for the repeated measures ANOVA comparing the different muscle activation levels of each of the coordinated movements for the beginning (0–24%), middle (25–74%), end (75–100%), and whole (0–100%) sections of the rolling movement. The ANOVA results where significance was found are bold and the significant results of the Dunnett’s T3 post-hoc test are shown below the line. Significance marked with an asterisk (*) indicates no post-hoc significance was found.

	Beginning	Middle	End	Whole
ES	p=0.36	p<0.005 D > F (p=0.01)	p=0.12	p=0.01*
HAM	p=0.31	p=0.04*	p=0.38	p=0.12
AB	p=0.02*	p<0.005 B < C (p<0.005) B < F (p=0.04) C > E (p<0.005) D < E (p=0.03) E < F (p<0.005)	p<0.005 C > E (p<0.005) D > E (p=0.02) E < F (p<0.005)	p<0.005 B < C (p<0.005) C > D (p=0.04) C > E (p<0.005) D > E (p=0.01) E < F (p<0.005)
QUAD	p=0.01 B < F (p=0.04)	p=0.11	p=0.87	p=0.17

4. DISCUSSION

The purpose of this study was to determine the differences in muscle activity required during unassisted rolling in healthy infants for six previously defined coordinated movements (Kobayashi et al., 2016).

4.1. Coordinated Movements

All six previously defined coordinated movements were utilized by infants in this study. Previous research found that younger infants (5–7 months) tend to use coordinated movements A, B, C, and D which feature stationary limbs more often, while older infants (8–10 months) begin to utilize coordinated movements E and F as well. This is because younger infants rely on the stability provided by stationary limbs (Kobayashi et al., 2016). The limited age range of infant participants (4.6–7.6 months) explains why there was no statistical difference between the coordinated movements and age.

4.2. Muscle Activation

All coordinated movements began with every muscle firing at the same time. After that initial period, the muscle activations varied for the middle and end of the rolling movement based on limb coordination of the six previously defined coordinated movements (Kobayashi et al., 2016), confirming our hypothesis.

The two coordinated movements which feature trunk and hip flexion (C and F) occurred most often in our study. Coordinated movement F has no stationary limbs, meaning all limbs move to achieve the roll together. Movement C is similar, but the ipsilateral arm is stationary. The moderate to high abdominal and quadricep muscle activity throughout these rolling movements corresponds with the trunk and hip flexion. For the beginning and middle of the rolling movement, the abdominal muscle activation was higher for coordinated movement C compared to F, showing that while these movements do have similar abdominal muscle activation overall, the timing is different throughout the roll. This finding indicates that the use of stationary limbs influences abdominal muscle activation.

Two coordinated movements (B and E) feature axial rotation of the torso relative to the pelvis. The lower body movements are the same but coordinated movement B uses only the contralateral arm to initiate the roll and movement E uses both arms to initiate the roll. For the middle and end of these movements, the erector spinae muscles exhibited moderate to high muscle activation which facilitates the axial rotation needed to achieve these movement types (Mayer et al., 2012) as well as the push-off mechanism that occurs during these movements (Kobayashi et al., 2021). Coordinated movement B exhibited higher erector spinae muscle activation throughout the rolling movement compared to movement E. Once again, this indicates that stationary limbs can influence muscle activation. The increase in muscle activation of the erector spinae and abdominal muscles when using stationary limbs indicates that infants rely more heavily on these muscle groups as they are learning to achieve a roll. As infants develop, the muscle activation threshold required to initiate a roll may decrease for the erector spinae and abdominal muscles and increase for other muscle groups as they begin to utilize fewer stationary limbs.

Coordinated movement D was the third most prevalent roll type, featuring axial rotation of the torso relative to the pelvis as well as hip flexion. All muscles were working at a moderate level of muscle activity throughout the middle and end of the coordinated movement, indicating that infants must use many muscles synchronously to achieve this roll type. Coordinated movement D also exhibited significantly higher erector spinae muscle activation for the middle of the rolling movement compared to movement F and significantly lower abdominal muscle activation for the whole rolling movement compared to movement C. Movements C and F also utilize hip flexion, but since movement D incorporates axial rotation of the torso relative to the pelvis, the combination of this rotation and hip flexion increased the erector spinae muscle activation. Additionally, the increased erector spinae muscle activation is indicative of the push-off mechanism utilized, similar to movements B and E.

For coordinated movement A, all muscle activity decreased for the end of the rolling movement except for the quadricep muscles, which increased to the high muscle activation region. This indicates that the stationary ipsilateral leg is performing a mechanism similar to the spinal extensor phase that was first described as pre-rolling by (McGraw, 1941). This pre-rolling movement is demonstrated when the infant extends their neck dorsally by placing their foot against the floor surface causing spinal extension and pelvic thrust. This phase of rolling occurs when an infant is just learning how to roll, indicating that this mechanism may be used to provide stability to inexperienced rollers and younger infants. While this study provides a baseline for rolling on a firm flat surface, different mechanical environments and product design features should be considered as they may facilitate or hinder a push-off mechanism, potentially allowing infants to achieve a roll prematurely, before they have achieved a roll on a flat surface.

In addition to providing a baseline of infant muscle utilization during rolling for future juvenile products research, characterizing muscle activation patterns between the six different coordinated movements is also useful information for clinicians. The assessment of movement patterns and muscle activity during rolling could be an additional evaluation component in the early identification of movement disorders like cerebral palsy and hypertonia (Bennett, 1999; Burns et al., 1989; Harris and Heriza, 1987). With a more complete understanding of the muscles which healthy infants are using to achieve a roll, physical therapists will also have insight into what muscles may be lacking strength and coordination in infants with various motor delays.

4.3. Considerations

Our data collection occurred throughout the COVID-19 pandemic (October 2020-June 2021), meaning some infants had minimal experience outside of their homes or with strangers. The infants’ inexperience with new and unfamiliar environments, as well as the sensors placed on them, may have impacted the coordinated movements chosen and be different than ones executed in the comfort of their own homes. There are also inherent limitations in using surface EMG sensors on adults and children. However, the sensors chosen have been used in other infant studies without hindering normal infant movement, and our experimental design allowed infants to serve as their own controls, limiting the variability sometimes associated with EMG placement (Mannen et al., 2019; Price et al., 2018; Siddicky et al., 2020, 2021; Solopova et al., 2019; Wang et al., 2020, 2021). Lastly, the muscle groups we chose to explore in this study were not comprehensive but were selected based on the previously defined coordinated movements during rolling (Kobayashi et al., 2016; McGraw, 1941).

Future studies should expand the age range to determine how rolling strategies and muscle activation changes with age and development. Additionally, gathering a larger sample size of infants rolling on a flat surface would allow for a powered comparison of contralateral versus ipsilateral muscle utilization and provide a more complete look at muscle activity during the six coordinated movements of rolling. Since this study only analyzed rolling on a firm flat surface, future work should also seek to quantify how rolling is achieved in different mechanical environments.

4.4. Conclusion

This study is the first that we know of to explore the muscle activity required during the various coordinated movements of infant rolling on a flat surface. For coordinated movements that use axial rotation of the torso relative to the pelvis to complete the roll, we measured higher levels of erector spinae muscle activation. For coordinated movements where trunk and hip flexion are required, higher levels of abdominal muscle activation were measured. Understanding how infants coordinate their movements and use their muscles during rolling on a firm flat surface is crucial to providing more insight into a key stage of infant motor development and providing a baseline for future research into the effects of mechanical environment on infant rolling.