Competition for limited neural resources in older adults leads to greater asymmetry of bilateral movements than in young adults

Abstract

We previously demonstrated that lateralization in the neural control of predictive and impedance mechanisms is reflected by interlimb differences in control of bilateral tasks. Aging has been shown to reduce lateralization during unilateral performance, presumably due to greater recruitment of the ipsilateral hemisphere. We now hypothesize that aging-related reduction in the efficiency of neural resources should produce greater behavioral asymmetry during bilateral actions that require hemispheric specialization for each arm. This is because simultaneous control of dominant and nondominant arm function should induce competition for hemisphere-specific resources. To test this hypothesis, we now examine the effect of aging (young, n = 20; old, n = 20) on performance of a mechanically coupled task, in which one arm reaches toward targets while the other arm stabilizes against a spring that connects the two arms. Results indicate better dominant arm reaching performance and better nondominant arm stabilizing performance for both groups. Most notably, limb and joint compliance was lower in the dominant arm, leading to dominant arm deficits in stabilizing performance. Group analysis indicated that older adults showed substantially greater asymmetry in stabilizing against the spring load than did the younger adults. We propose that competition for limited neural resources in older adults is associated with reduced contributions of right hemisphere mechanisms to right-dominant arm stabilizing performance, and thus to greater asymmetry of performance.

NEW & NOTEWORTHY We provide evidence for greater asymmetry of interlimb differences in bilateral coordination for stabilizing and preserved asymmetry of reaching with aging. These results provide the first evidence for increased lateralization with aging within the context of a complementary bilateral task.

INTRODUCTION

Age-related decline in arm and hand motor function is common and includes loss of strength (Baumgartner et al. 1998), dexterity (Louis et al. 2005), coordination difficulty (Seidler et al. 2002), and slowing of movements (Buckles 1993). A decline in motor performance has also been identified during bilateral arm movements, which make up the majority of everyday activities (Bailey et al. 2015; Kilbreath and Heard 2005). Compared with that in young adults, bilateral coordination is less accurate and more variable in older adults (Seidler et al. 2010), with deficits observed during antiphase (vs. in-phase) bilateral tasks (Bangert et al. 2010; Lee et al. 2002; Swinnen 1998; Wishart et al. 2000). In older adults, difficulty performing bilateral tasks has been linked to cognitive and motor age-related reductions in lateralized asymmetry (Collins and Mohr 2013; Paizis et al. 2014). Nevertheless, although evidence of age-related reductions in performance asymmetry between the dominant and nondominant arm has been identified, these studies are limited to investigations of unilateral arm movements (Przybyla et al. 2011; Wang et al. 2011). These tasks, while important to an initial understanding of age effects, fail to provide an understanding of everyday complementary tasks where one hand stabilizes and one hand manipulates, such as holding a baguette with one hand and slicing it with the other. Furthermore, it is unclear if age-related reductions in asymmetry contribute to deficits seen during bilateral tasks.

We previously demonstrated that lateralization of neural control of predictive and impedance mechanisms is reflected by interlimb differences in controlling complementary bilateral tasks in young adults (Woytowicz et al. 2018). More specifically, the prior study demonstrated better kinematic performance of the right, dominant arm for reaching and better kinematic and dynamic performance of the left, nondominant arm for stabilizing. These results provided the first evidence in young adults for lateralized arm-hemisphere specializations for reaching and stabilizing during a mechanically coupled complementary bilateral task.

Aging has been shown to reduce lateralization of both cognitive (Berlingeri et al. 2013; Cabeza 2002) and motor (Fling and Seidler 2012) tasks, although the effect of aging on motor lateralization is still unclear, with other studies demonstrating no change in lateralization (Chua et al. 1995; Pan and Van Gemmert 2013). Studies that have demonstrated age-related reductions in motor lateralization include hand dominance attenuation with a shift toward ambidexterity (Kalisch et al. 2006), reduced laterality during imagined actions (Paizis et al. 2014), reduced asymmetry during multidirectional reaching movements (Przybyla et al. 2011), and reduction in interlimb transfer of a visuomotor adaptation during reaching (Wang et al. 2011). In contrast, others have failed to identify age-related changes in hand laterality when tasks were broken down into preparation and execution movement stages (Chua et al. 1995), and preservation of interlimb transfer occurred during a multidirectional drawing task (Pan and Van Gemmert 2013). In addition to these conflicting results, it is important to note that the evidence is based solely on unimanual motor tasks. Therefore, it is not yet known whether or how age-related changes in motor lateralization will affect motor performance during a bilateral task.

The present study aimed to investigate whether aging affects motor performance and lateralization asymmetry during a bimanual mechanically coupled task. The task was modeled after common everyday tasks, such as slicing bread or hammering nails, in which one hand stabilizes against forces applied to an object by the other hand. We first reasoned that since aging reduces the efficiency of motor control resources in each hemisphere, overall performance for each arm should also be decreased. This hypothesis is supported by previous studies of unilateral movements in elderly participants (Kalisch et al. 2006; Paizis et al. 2014; Przybyla et al. 2011; Wang et al. 2011). Furthermore, overactivation of brain areas in older adults has been observed for a wide range of processes, including motor control at lower levels of task demand or to achieve similar performance to young adults (Berlingeri et al. 2013; Heuninckx et al. 2005, 2008; Reuter-Lorenz and Cappell 2008; Reuter-Lorenz and Lustig 2005). This is thought to be compensatory and to limit the neural resources available as task demands increase. For example, in a prior study, as the complexity of bilateral tasks increased, the patterns of brain activity increased and extended into brain areas for sensory and cognitive function, suggesting the processing was less automatic in older adults and was associated with decreased motor performance (Heuninckx et al. 2005). Second, the age-related reduction in neural resource efficiency should also lead to greater performance asymmetry during bilateral tasks in older compared with young adults. More specifically, we predicted that, compared with young adults, older adults would demonstrate 1) a greater reaching advantage for the dominant arm and 2) a greater stabilizing advantage for the nondominant arm, independent of overall performance decreases for both arms. This hypothesis is based on our bihemispheric model of motor lateralization, in which each hemisphere differentially contributes to the control of each arm. With increased interhemispheric connectivity that occurs with aging (Heuninckx et al. 2008), competition for the limited intrahemispheric resources will be stronger contralaterally than ipsilaterally. As such, the processing demands for performing a nonspecialized role (i.e., nondominant arm reaching and dominant arm stabilizing) will be even greater and less efficient in older adults, thereby increasing the performance asymmetry between each arm. To test this hypothesis, we compared arm-hemisphere specializations in reaching and stabilizing performance during a complementary bilateral task in young vs. older adults.

METHODS

Participants.

Twenty right-handed adults aged between 21 and 35 yr (“young adults” group; 27.4 ± 2.8 yr, mean ± SD; 10 men and 10 women) and 20 right-handed adults, 65 yr and older (“older adults” group; 73.4 ± 6.6 yr; 10 men and 10 women), participated in this study. The 13-item Edinburgh Handedness Inventory was used to confirm that all participants were right-hand dominant (i.e., laterality quotient of 80 or greater; Oldfield 1971). Handedness scores were not significantly different between young and older participant groups. Exclusion criteria included any neurological condition such as stroke, Parkinson’s disease, or multiple sclerosis; significant musculoskeletal impairments including any upper extremity bone fractures occurring within the past 5 yr and/or soft tissue injuries that would affect performance of the motor task; and musicians, defined as anyone who plays or played a musical instrument regularly (>1 h/wk) for an extended period of time (>1 yr). All participants gave written informed consent before testing and were paid for their participation. The Institutional Review Board the University of Maryland School of Medicine and the Institutional Review Board of the Penn State College of Medicine approved all experimental procedures. Results from the group of young adults were published previously as a separate behavioral study (Woytowicz et al. 2018). While all data were collected at the Maryland School of Medicine, deidentified data were shared between sites for analysis.

Experimental design.

The testing paradigm was implemented using KineReach, a custom virtual reality motion tracking system developed at Penn State by R. Sainburg. Participants sat facing a projection screen with both arms below the screen and supported on air cushion sleds by continuous pressurized airflow to maintain the limbs against gravity and reduce friction. A brace was used to immobilize all joints distal to the elbow, and trunk and scapula movements were restricted with the use of a chest restraint, but motion of the trunk and scapula were reflected as recorded motion of the shoulder (acromion). Position and orientation of the forearm and upper arm segments were recorded using three-dimensional 10-degree-of-freedom motion of each arm at 120 Hz with Flock of Birds motion sensors (Ascension Technology). The positions of the hand, elbow, wrist, and shoulder were computed using two Flock of Birds markers per arm, with the X-Y plane parallel to the tabletop and digitization to the proximal interphalangeal joint of the second metacarpal, space between the third and fourth metacarpophalangeal joints, styloid processes of the ulna and radius, medial and lateral epicondyle of the elbow joint, and acromion of both arms. Computed X-Y coordinates of the fingertip were used to define the projected cursor position. Real-time cursor (finger) position, a start circle, and a target circle were displayed on the projection screen.

As described in a prior study (Woytowicz et al. 2018), all participants completed two tasks: 1) right hand reaching and left hand stabilizing, and 2) left hand reaching and right hand stabilizing, with half the participants completing these tasks in the reverse order (pseudorandomly divided to ensure equal division of male and female participants). The task was performed with a spring affixed to the arm sleds (versus the arm or hand to avoid potential haptic feedback effects). Both the left- and right-hand cursors (diameter = 0.015 m) were displaced 0.127 m medially from the corresponding hand. This produced spring torques of ~1 and 4 N·m at the elbow and shoulder joints, respectively. To begin each trial, both hand cursors were required to be within the start circle (diameter = 0.03 m). Seven target circles (diameter = 0.03 m) were located radially 0.2 m from the start circle, one target directly anterior to the start circle and three targets 15, 30, and 45 degrees medially and laterally. Each task condition consisted of 10 testing blocks of seven trials with one trial to each target direction in randomized order (70 total trials for each task condition). For a visual illustration of the experimental task and target display, see Fig. 1 in Woytowicz et al. (2018).

To ensure participants moved fast enough to elicit potential performance differences between arms and within a similar range between arms and participants, they were asked to reach at a minimum speed of 0.5 m/s, as accurately as possible. Simultaneously, participants were asked to maintain their stabilizing hand position within the start circle. During each trial, current finger position was displayed (cursor). After each trial, visual feedback regarding movement speed was provided with a horizontal speed bar displayed at the top of the screen. The length of the speed bar indicated whether the participants needed to increase the speed in the following trial). Visual feedback of task performance was also given at the end of every trial in the form of hand-path displays for both the right and left hands. For motivational purposes, when participants attained the minimum of 0.5 m/s or higher, they were awarded points based on how closely they reached the center of the target.

Data processing and analysis.

All kinematic, kinetic, and dynamic analyses were processed with custom programs written in Igor Pro (version 6.37; WaveMetrics) and were identical to the analysis methods described previously. (For more details see kinematic analysis, inverse dynamics, and mechanical compliance methods subsections of Woytowicz et al. 2018). The data were low-pass filtered at 8 Hz with a third-order dual-pass Butterworth filter before differentiation to obtain velocity and acceleration profiles. Movement onset and offset were defined using the reaching arm. The start of each reach was defined as the first minimum in tangential velocity that was under 8% of the maximum velocity for that trial. The end of each reach was defined as the first minimum following peak velocity that was below 8% of maximum velocity. Movement onset and offset of the stabilizing hand were then defined using these reaching movement times.

For the inverse dynamics, the terms in the equations of motion were partitioned into four main components to calculate 1) muscle torque, which estimates rotational forces from muscle contraction; 2) interaction torque, representing forces imposed by the movement of other limb segments; 3) net torque, the sum of the muscle and interaction torques (Bagesteiro and Sainburg 2002); and 4) spring torque. Interaction torque generated at the elbow resulted primarily from shoulder rotational accelerations and velocities, as well as linear accelerations of the shoulder. Limb segment inertia, center of mass location, and mass were computed using each participant’s body weight and limb segment lengths (Winter 1990). To characterize the contributions of muscle and interaction torque to stabilization of each joint against the spring load, we quantified muscle and interaction torques at the time of peak net torque, normalized by the net torque magnitude. Thus a larger value reflected greater contribution of either muscle or interaction torque to the net torque, which is directly proportional to joint acceleration. See appendix for equations of motion and components of joint torques (muscle torque, interaction torque, net torque, and spring torque) at each joint. For an illustration demonstrating the configuration of the spring and the associated force vectors, see Fig. 2 in Woytowicz et al. (2018).

Primary performance measures.

Primary kinematic measures included deviation from linearity and end error to compare reaching arm performance, and hand displacement and maximum acceleration to compare stabilizing arm performance. Lower values of both deviation from linearity and final position error indicate better reaching performance, and lower values of both hand displacement and maximum acceleration indicate better stabilizing performance.

To further characterize stabilizing performance, compliance (the tendency to move when a force is applied) was calculated at the hand and the stabilizing arm joints. Greater compliance is indicative of less stiffness (i.e., ineffective impedance control), given that compliance is the inverse of stiffness and stability of the arm is achieved by increasing stiffness (i.e., decreasing compliance) with increased torque (McIntyre et al. 1996). As a measure of end-point compliance, hand displacement was divided by the peak spring force. At the elbow and shoulder, joint displacement was regressed by the imposed spring torque, with the slope of this relationship defined as joint compliance.

Assessment of motor and cognitive function.

In addition to completing the experimental task, the motor and cognitive functions of all participants were also assessed. Assessments of motor function included the Purdue Pegboard Test (PPT) of manual dexterity (Tiffin and Asher 1948) and maximum grip strength. Maximum grip strength was measured with a BASELINE hydraulic handheld dynamometer using standardized instructions and positioning of the arm (seated with shoulder adducted and neutrally rotated, elbow flexed at 90°, and the forearm and wrist in neutral position), and using the average of two trials for both the left and right hand (Fess and Moran 1981; Innes 1999; Mathiowetz et al. 1985; Roush et al. 2017).

Cognitive function was assessed using the Test of Everyday Attention (TEA) (Robertson et al. 1996). Participants completed the Map Search, Elevator Counting, Elevator Counting with Distraction, Visual Elevator, Telephone Search, and Telephone Search Dual Task subtests of the TEA. The subtests were modeled after “real-life” activities and measured different types of attentional function. The Map Search and Telephone Search subtests provided measures of selective attention. Sustained attention was measured using the Elevator Counting subtest. The Elevator Counting with Distraction subtest provided a measure of working memory. Attentional switching ability was measured by the Visual Elevator subtest. Finally, the Telephone Search Dual Task provided a measure of divided attention. The Telephone Search subtest was required to score the performance of the Telephone Search Dual Task subtest; therefore, the Map Search subtest was used as our primary measure of selective attention. Raw scores were also converted to age-normalized scaled scores for all subtests.

Statistical analysis.

To ensure participants were familiarized with the task, and to control for any potential effects of interlimb transfer, we limited all analyses to steady-state performance. To ensure that we were including only participants’ steady-state performance, we calculated the mean coefficient of variation (CV) for peak velocity per testing block of all testing sessions (Sainburg et al. 2016). In addition, to control for any initial effects of force adaptation due to the spring, we calculated CV for end error of the second task condition performed with the spring. Each of the testing sessions consisted of 70 trials, separated into 10 blocks of 7 trials. We statistically assessed steady-state performance by using a 2 (age group: young, old) by 2 (arm: left, right) by 10 (block) mixed-factor ANOVA. The first blocks with highest CV were then excluded, steady-state performance was confirmed with a subsequent mixed factor ANOVA including only the remaining blocks, and dependent variables were then collapsed across these remaining blocks for further analysis.

Given the relationship between movement velocity and accuracy (Fitts 1954), before testing the primary hypothesis, we needed to first ensure that peak movement velocity of the reaching arm was comparable for both arms. To do so, we performed a mixed-factor ANOVA on mean peak velocity data with arm (right, left) and target direction (medial, center, lateral) as within-participant factors and age group (young, old) as a between-participant factor. We expected only a main effect of target direction, but no effect of arm, thereby indicating comparable left and right reaching arm peak velocities.

Since end-point stiffness varies based on arm orientation and force direction (Perreault et al. 2002), we grouped the trials with reaching targets located medial, center, and lateral with respect to the start position of the reaching arm. To compare reaching kinematic performance and stabilizing kinematic and dynamic performance, repeated-measure mixed-model ANOVAs with within-group factors of arm and target direction and a between-group factor of age group were used.

Finally, we compared assessments of motor and cognitive functions between groups and assessed the relationship between kinematic performance and functional outcomes. Interlimb asymmetry in kinematics was calculated as the ratio between left and right arm performance. Deviation from linearity was used as the measure of reaching performance, and end-point compliance was used as the measure of stabilizing performance. Given that higher values of both performance measures were indicative of worse performance, an asymmetry ratio >1 indicated greater left-hand error, while a ratio <1 indicated greater right-hand error. Functional outcomes included right and left maximum grip strength; the Map Search, Elevator Counting, Elevator Counting with Distraction, Visual Elevator, and Telephone Search Dual Task subtests of the TEA; and the right hand, left hand, both hands, and assembly subtests of the PPT.

To compare left- and right-hand maximum grip strength, a repeated-measure mixed-model ANOVA with a within-group factor of arm and between-group factor of age group were used. To examine whether differences in strength between the dominant and nondominant hand might have contributed to interlimb asymmetry of reaching and stabilizing performance, we assessed the relationship between interlimb asymmetry and maximum grip strength asymmetry using a Pearson’s correlation analysis, which was calculated as the ratio between maximal grip strength of the left and right arm. To compare manual dexterity, the right hand, left hand, both hands, and assembly subtests of the PPT were compared between age groups using an independent t test.

Given the high cognitive demands of bilateral coordination (Swinnen and Wenderoth 2004), scaled scores of attention were compared between age groups using an independent t test, to ensure both groups had similar age-normalized attentional function (excluding Elevator Counting, which is scored categorically as normal, possibly abnormal, and abnormal). Furthermore, given prior evidence illustrating reduced cognitive lateralization with aging (Cabeza 2002), Pearson’s correlation analyses were used to assess the relationship between raw measures of attention and measures of reaching and stabilizing interlimb asymmetry.

All statistical analyses were completed using JMP software (SAS Institute Inc., Cary, NC).

RESULTS

Steady-state performance.

Reaching velocity CV of the spring condition was highest during the initial two blocks, with greater differences in the old adults, as illustrated in Fig. 1 and reflected by a significant main effect of block [F(9,342) = 3.66, P = 0.0002] and age group [F(1,38) = 18.19, P = 0.0001] but no significant interaction. However, for the remainder of the session (blocks 3–10), there was no significant main effect of block [F(7,266) = 1.61, P = 0.1325] for the CV of reaching velocity. There was no effect of block for the CV of reaching arm end error between conditions [F(9,324) = 1.11, P = 0.3529] and no effect of age group [F(1,36) = 3.59, P = 0.0660] or interaction. Therefore, remaining results were analyzed using the steady-state performance observed during blocks 3–10.

Fig. 1.

Steady-state performance. Mean (SE) coefficient of variation (CV) of end error is plotted for both left (dashed line) and right (solid line) reaching arms, for both the young (left) and older (right) adults. Each data point shows the average of every 7 trials, or 1 block, across the group. Reaching velocity CV of the spring condition was highest during the initial 2 blocks, with greater differences in the old adults [main effect of block: F(9,342) = 3.66, P = 0.0002, and age group: F(1,38) = 18.19, P = 0.0001]. Blocks 1 and 2 (gray-shaded areas) were removed, and blocks 3–10 reflect the steady-state performance [no significant effect of block: F(7,266) = 1.61, P = 0.1325].

As expected, movement velocity of the reaching arm did vary by target [F(2,76) = 281.77, P < 0.0001] but did not vary between the arms [F(1,38) = 0.34, P = 0.5312], and there was no interaction. Therefore, target was included as a within-subject factor for the subsequent results. Velocity also varied by age group [F(1,38) = 18.93, P < 0.0001]. As illustrated in Fig. 2, young adults reached faster than older adults, although there were similar within-group differences between the targets, with the fastest reaches in the lateral direction.

Fig. 2.

Reaching velocity. Mean (SE) movement velocity of the reaching arm is plotted for both left (dashed line) and right (solid line) reaching arms, for both the young (left) and older (right) adults, across the 3 target group locations. Movement velocity varied by target [F(2,76) = 281.77, P < 0.0001] and by age group [F(1,38) = 18.93, P < 0.0001] but did not vary between the arms [F(1,38) = 0.34, P = 0.5312]. Max., maximum.

As shown in Fig. 3A, the left reaching arm illustrated significantly greater deviation from linearity compared with the right arm [F(1,38) = 32.54, P < 0.0001], which varied by movement direction [F(2,76) = 75.60, P < 0.0001]. There was also a significant effect of age group, in which older adults’ reaching performance demonstrated significantly greater deviation from linearity [F(1,38) = 19.55, P < 0.0001], especially for more medial targets, compared with that of young adults, as reflected by a significant age-by-target interaction [F(2,76) = 5.18, P = 0.0078]. Figure 3A shows that end error also varied between the arms [F(1,38) = 16.59, P = 0.0002], which also varied by movement direction [F(2,76) = 4.07 P = 0.0209], but there was no significant effect of age group [F(1,38) = 1.99, P = 0.1656] or interaction. Thus, as in investigations of unilateral asymmetries, the right arm illustrated straighter reaching trajectories during a complementary bilateral task with a mechanical interaction between the arms and also demonstrated superior performance regarding accuracy of the final arm position at the target.

Fig. 3.

Kinematic performance. Mean (SE) kinematic performance measures are plotted across the 3 target group locations, for both young and older adults. A: reaching arm kinematics of deviation from linearity (left) and end error (right). The left reaching arm illustrated greater deviation from linearity compared with the right arm [F(1,38) = 32.54, P < 0.0001], which varied by movement direction [F(2,76) = 75.60, P < 0.0001]. Older adults’ reaching performance demonstrated greater deviation from linearity [F(1,38) = 19.55, P < 0.0001], especially for more medial targets compared with young adults [age by target interaction: F(2,76) = 5.18, P = 0.0078]. End error varied between the arms [F(1,38) = 16.59, P = 0.0002] and by movement direction [F(2,76) = 4.07, P = 0.0209]. B: stabilizing arm kinematics of hand displacement (left) and maximum acceleration (right). The right arm moved [F(1,38) = 33.79, P < 0.0001] and accelerated [F(1,38) = 15.91, P = 0.0003] more than the left arm, and hand displacement increased with age [F(1,38) = 8.19, P = 0.0068]. There was greater performance asymmetry between the left and right arm stabilizing performance in older adults [age by hand interaction for hand displacement: F(1,38) = 6.51, P = 0.0148, and acceleration: F(1,38) = 4.49, P = 0.0407]. a.u., Arbitrary units.

Stabilizing arm.

When stabilizing, the right arm moved [F(1,38) = 33.79, P < 0.0001] and accelerated [F(1,38) = 15.91, P = 0.0003] more than the left arm, as illustrated in Fig. 3B. Thus, when a torque is applied by a mechanical interaction between the arms, the results demonstrate an advantage of the left arm to remain stable (less hand displacement) compared with the right arm. Overall stabilizing performance reduced with age, reflected by a significant main effect of age for hand displacement [F(1,38) = 8.19, P = 0.0068] in our mixed-factor ANOVA but no interaction. In addition, there was greater performance asymmetry between the left- and right-arm stabilizing performance in older adults, as illustrated in Fig. 3B and reflected by a significant age-by-hand interaction for both hand displacement [F(1,38) = 6.51, P = 0.0148] and acceleration [F(1,38) = 4.49, P = 0.0407].

We also compared the performance of the stabilizing arm between groups by normalizing hand displacement by peak spring force as a measure of end-point compliance. As illustrated in Fig. 4A, this measure of end-point compliance was greater for the right arm vs. the left arm, with a significant main effect of arm [F(1,38) = 30.97, P < 0.0001], and there was also a main effect of age [F(1,38) = 10.64, P = 0.0023], with older adults demonstrating overall greater end-point compliance compared with young adults. In addition, there was a significant age-by-hand interaction [F(1,38) = 6.09, P = 0.0182], with greater performance asymmetry between left- and right-arm end-point compliance in older adults. Furthermore, Fig. 4B illustrates that the right elbow was more compliant compared with the left, with a significant main effect for arm [F(1,38) = 8.68, P = 0.0055]. Figure 4C illustrates that this interlimb difference in compliance was not found at the shoulder [F(1,38) = 0.08, P = 0.7819], although there does appear to be a trend for the right shoulder to be more compliant than the left for the center and lateral targets. However, there was no effect of age on compliance at the elbow [F(1,38) = 0.33, P = 0.5667] or at the shoulder [F(1,38) = 0.64, P = 0.4272], and no interaction. Thus the right arm has a greater tendency to move when a force is imposed, which appears to occur primarily due to interlimb control differences at the elbow for both young and older adults. However, the differential effects at the joints are likely related to the geometry of the task. Because the spring connected the right and left arm supports, the primary action of the force was perpendicular to the forearm, thus producing greater torque at the elbow than at the shoulder.

Fig. 4.

Stabilizing arm compliance. Mean (SE) compliance measures are plotted across the 3 target group locations, for both young and older adults. A: end-point compliance was greater for the right arm vs. the left arm [F(1,38) = 30.97, P < 0.0001] and was greater in older vs. young adults [F(1,38) = 10.64, P = 0.0023]. There was greater performance asymmetry between left and right arm end-point compliance in older adults [age by hand interaction: F(1,38) = 6.09, P = 0.0182]. B: elbow compliance was greater in the right vs. left arm [F(1,38) = 8.68, P = 0.0055] but did not vary by age [F(1,38) = 0.33, P = 0.5667]. C: shoulder compliance did not vary between the arms [F(1,38) = 0.08, P = 0.7819] or by age group [F(1,38) = 0.64, P = 0.4272].

Taken together with our kinematic results, our results provide convergent evidence that 1) specialization of the right (dominant) arm for predictive control and left (nondominant) arm for impedance control in the bilateral task are preserved with aging; and 2) interlimb differences for stabilizing were greater in older than in younger adults, indicating an increase rather than a decrease in this aspect of motor lateralization.

Interlimb asymmetry related to cognitive and motor functions.

Mean values of reaching and stabilizing asymmetry ratios, left and right maximum grip force, the Map Search, Elevator Counting, Elevator Counting with Distraction, Visual Elevator, and Telephone Search Dual Task subtests of the TEA, and the right hand, left hand, both hands, and assembly subtests of the PPT are presented in Table 1. For maximum grip strength, there was a main effect of hand [F(1,38) = 16.45, P = 0.0002] and age group [F(1,38) = 8.13, P = 0.0070], and no hand-by-age group interaction [F(1,38) = 0.23, P = 0.6307], indicating that differences in strength between the hands did not increase with age. Results of the Pearson correlation indicated that there was no significant association between maximum grip strength asymmetry and reaching asymmetry [R²(39) = 0.04, P = 0.2341] or stabilizing asymmetry [R²(39) = 0.01, P = 0.6293], indicating that differences in interlimb asymmetry occurred independently from differences in strength between the dominant and nondominant hand. The independent samples t tests found significant differences in right hand [t(37.2) = 6.61, P < 0.0001], left hand [t(33.5) = 5.96, P < 0.0001], both hands [t(34.4) = 5.45, P < 0.0001], and assembly [t(32.1) = 9.69, P < 0.0001] PPT scores between groups, indicating a significant reduction in dexterity in older compared with young adults.

Table 1.

Cognitive and motor function summary

	Age Group
	Young adults	Older adults
Reach asymmetry (L:R)	1.27 (0.21)	1.18 (0.24)
Stabilize asymmetry ratio (L:R)	0.83 (0.32)	0.72 (0.20)
L maximum grip strength, kg	41.70 (10.62)	32.46 (10.21)
R maximum grip strength, kg	43.93 (11.17)	34.23 (10.42)
Grip strength asymmetry (L:R)	0.95 (0.07)	0.95 (0.09)
Purdue Pegboard Test
Right hand	17.2 (2.2)	12.1 (2.6)
Left hand	15.8 (1.8)	11.4 (2.7)
Both hands	13.5 (1.9)	9.3 (2.7)
Assembly	43.3 (5.0)	23.2 (7.8)
TEA age-normalized scores
Map Search	9.1 (3.78)	9.9 (2.40)
Elevator Counting	N:17, PA:3, A:0	N:16, PA:2, A:2
Elevator Counting with Distraction	10.15 (2.87)	9.35 (2.64)
Visual Elevator	10.9 (2.81)	11.55 (3.10)
Telephone Search Dual Task	11.75 (4.28)	10.2 (3.14)
TEA raw scores
Map Search (no. found)	74.2 (4.70)	53.4 (14.64)
Elevator Counting (no. correct)	6.85 (0.37)	6.65 (0.81)
Elevator Counting with Distraction (no. correct)	8.15 (2.54)	7.4 (2.54)
Visual Elevator (no. correct)	8.55 (2.04)	8.1 (2.43)
Telephone Search Dual Task (time per target increase)	0.67 (1.41)	2.02 (2.72)

Values are means (SD) for 20 right-handed young adults between 21 and 35 yr of age (27.4 ± 2.8 yr) and 20 right-handed older adults, 65 yr and older (73.4 ± 6.6 yr). L, left; R, right; TEA, Test of Everyday Attention (for age-normalized Elevator Counting, N = normal, PA = possibly abnormal, and A = abnormal).

The independent samples t tests found no significant differences in Map Search [t(38) = −0.80, P = 0.4296], Elevator Counting with Distraction [t(38) = 0.92, P = 0.3648], Visual Elevator [t(38) = −0.69, P = 0.4915], and Telephone Search Dual Task [t(38) = 1.31, P = 0.1993] scaled scores between groups, indicating both older and younger adults had similar age-normalized attentional function. However, independent samples t tests of raw TEA task scores found significant differences in Map Search [t(38) = 6.05, P < 0.0001] and Telephone Search Dual Task [t(38) = −1.97, P = 0.0279], with no differences in Elevator Counting with Distraction [t(38) = 0.93, P = 0.3565] and Visual Elevator [t(38) = 0.63, P = 0.5291] between groups, indicating a decline in selective and divided attention function in the older adults compared with the young adults.

Results of the Pearson correlation indicated that there were no significant associations between raw Map Search scores and reaching [R²(39) = 0.03, P = 0.3023] or stabilizing asymmetry [R²(39) = 0.02, P = 0.3341], raw Elevator Counting and reaching [R²(39) = 0.02, P = 0.4388] or stabilizing asymmetry [R²(39) = 0.00, P = 0.8306], raw Elevator Counting with Distraction and reaching [R²(39) = 0.00, P = 0.9714] or stabilizing asymmetry [R²(39) = 0.00, P = 0.6910], raw Visual Elevator scores and reaching [R²(39) = 0.00, P = 0.9244) or stabilizing asymmetry [R²(39) = 0.07, P = 0.1041], and raw Telephone Search Dual Task scores and reaching [R²(39) = 0.03, P = 0.2650] or stabilizing asymmetry [R²(39) = 0.05, P = 0.1800], indicating that differences in interlimb asymmetry occurred independent of selective attention, sustained attention, working memory, attentional switching, and divided attention abilities, respectively.

DISCUSSION

The current study investigated whether age-related changes in arm-hemisphere specializations occur during a complementary bilateral task. We found that absolute performance decreased, asymmetric reaching specializations were preserved, and asymmetry in stabilizing increased with age. Increased asymmetry with aging was independent from differences in grip strength between the dominant and nondominant hands and from attention measures despite reduced maximum grip strength and selective and divided attention with age.

Based on our bihemispheric model of motor control (Sainburg 2005, 2014), we hypothesized that each arm utilizes motor control resources from each hemisphere, and contributions from the contralateral hemisphere are greater due to more extensive connectivity, leading to motor asymmetries. When intrahemispheric resources are reduced and become less efficient, competition from the ipsilateral controller should be disproportionately reduced, due to greater competition from the contralateral controller. Thus we predicted that aging should increase performance asymmetries during bilateral tasks, when both controllers are competing for resources from both hemispheres. Our results partially support this prediction, showing a nonsignificant increase in reaching performance asymmetries toward medial targets and significantly increased stabilizing performance asymmetries in older adults. These results contrast previous studies of unilateral reaching behavior, which illustrated decreased reaching asymmetry (Kalisch et al. 2006; Paizis et al. 2014; Przybyla et al. 2011; Wang et al. 2011). During unilateral tasks, there is no competition for motor control resources in the ipsilateral hemisphere; thus both hemispheres are available to support unilateral control. There are no previous studies of stabilizing performance in unilateral tasks to compare with our current results.

Age-related changes in performance.

Consistent with previous unilateral and bilateral literature, performance of both arms was reduced in older adults. For reaching movements, older adults were slower and less straight than young adults. For stabilizing performance, both hands of older adults demonstrated greater and faster displacement compared with young adults. Previous studies of bimanual performance have found the greatest age-related deficits during antiphase (vs. in-phase) bilateral tasks (Bangert et al. 2010; Lee et al. 2002; Swinnen 1998; Wishart et al. 2000). Given that greater deficits are found during nonsymmetrical vs. symmetrical actions (Collins and Mohr 2013; Paizis et al. 2014), it is possible that the performance deficits we observed would be comparable or greater than those found previously during antiphase tasks. Indeed, the older adults’ performance scores on the assembly subtest of the PPT, requiring nonsymmetric actions, were about half the performance scores of the young adults, demonstrating the largest difference between groups compared with the other PPT subtests, requiring unilateral and bilateral symmetric actions. However, without a direct comparison of in-phase and antiphase task performance to our results, we can only speculate on this point.

It has been well established that speed-accuracy tradeoff changes occur with aging such that elderly individuals move slower to achieve similar accuracy. This has previously been attributed to longer processing delays associated with reduced connectivity (Forstmann et al. 2011). Prior evidence for reduced unilateral motor lateralization was also associated with reduced reaching velocity (Przybyla et al. 2011). It is plausible that older participants in the prior study may not have moved fast enough to require the movement demands to elicit performance asymmetries. We attempted to control the reaching speed of all participants, although older adults still reached at a slower velocity compared with young adults but showed preserved reaching asymmetry. This suggests that the slower reaching velocity range was sufficient to require movement demands great enough to elicit asymmetric performance. These differences might increase if older adults were required to match the increased speed of younger adults or might decrease if younger adults were required to match the reduced speed of older adults. The advantage to controlling reaching velocity between age groups is that it may enable a more accurate comparison of age-related changes. However, given that older adults move slower than young adults in everyday life, controlling reaching velocity may also be a disadvantage for a functional interpretation of the results. Future research could test both populations using maximal or comfortable speeds.

Age-related changes in motor lateralization.

The present study provides the first evidence of increased motor performance asymmetry with age. As cited earlier, studies have demonstrated both reduced and preserved motor lateralization with aging, although evidence has been based on unilateral tasks. Our results may differ from these earlier studies because bilateral tasks require different interhemispheric processing than unilateral tasks, which also changes with aging. Increased reliance on interhemispheric connections with aging occurs via the corpus callosum (Zaidel and Iacoboni 2003), where the size and integrity of the corpus collosum reduces with age (Fling et al. 2011). During unilateral tasks, there is activation of both hemispheres (Derosière et al. 2014; Kawashima et al. 1994; Kim et al. 1993; Kobayashi et al. 2003; Remy et al. 1994; van Wijk et al. 2012; Verstynen et al. 2005), with greater ipsilateral recruitment during more complex tasks (Chen et al. 1997). With aging, bilateral recruitment for unilateral tasks increases (Heuninckx et al. 2005; Mattay et al. 2002), and at a structural level, reductions in white matter volume illustrate reduced structural evidence of motor lateralization (Koppelmans et al. 2015).

Furthermore, age-related declines in bilateral coordination are modulated by a combination of structural (white matter integrity) and functional neural changes (Fujiyama et al. 2016). In young adults, interhemispheric inhibitory interactions are essential for preventing interference from the opposite hemisphere during bilateral tasks (Goble et al. 2010; Rémy et al. 2008), while older adults utilize interhemispheric facilitation (Fling et al. 2011; Heitger et al. 2013). Greater white matter integrity of functionally organized corpus collosum subregions were associated with better performance of different behavioral tests of bilateral function, which suggests age-related changes to the corpus collosum and bilateral coordination are task specific and likely occur within the cortical regions connected to these pathways (Serbruyns et al. 2015). In addition, while increased interhemispheric connectivity may compensate for age-related structural and functional changes during unilateral and symmetric tasks, a declined ability to modulate these connections has been associated with deficits during more complex bilateral tasks. More specifically, this appears to occur between areas processing information for motor planning and monitoring (Fujiyama et al. 2016; Kiyama et al. 2014; Verstynen et al. 2005). Taken together, these findings suggest age-related changes to interhemispheric connections may impair sensorimotor processing important for the nonspecialized hemisphere during complex tasks, which increases interlimb asymmetry during complementary bilateral tasks.

Interlimb asymmetry and functional outcomes.

Prior laterality investigations have suggested that a stronger right hand was related to better performance on motor tasks assessing speed and visual-spatial skills (Cerone and McKeever 1999; Dellatolas et al. 2003). Our results demonstrated greater right-hand grip strength and superior right-arm reaching performance (vs. left arm) for both age groups but reduced absolute grip strength and slower movements in older compared with younger adults. However, relative differences between dominant and nondominant grip strength did not increase with age, indicating that increased stabilizing asymmetry in older adults was independent of differences in right vs. left strength and likely reflects changes in lateralization of motor control.

Previous researchers have suggested that increased cognitive demands of bilateral coordination (Swinnen and Wenderoth 2004) may magnify age-related bilateral coordination differences due to the known cognitive declines with aging (Bangert et al. 2010). The present study suggests that greater stabilizing asymmetry and preserved reaching asymmetry during bilateral coordination occurred independently of changes in attentional function with aging, as measured by the TEA subtasks. However, these measures of attention may not reflect attentional ability when both a cognitive and motor task are simultaneously performed and do not measure changes in attentional lateralization.

Comparison to cognitive model of reduced lateralization.

Cognitive aging investigations have provided the majority of previous evidence for reduced asymmetry and are conceptualized as the HAROLD model (hemispheric asymmetry reduction in older adults) (Cabeza 2002). Under this model, reduced lateralization of cognitive function was suggested to reflect compensation for age-related neurocognitive deficits (Cabeza et al. 1997) or, alternatively, age-related dedifferentiation of cognitive abilities (for review, see Li and Lindenberger 1999). Increased bilateral brain activity has been associated with enhanced performance of cognitive tasks, providing support for the compensation view of the HAROLD model (Reuter-Lorenz et al. 1999). For reaching, this model may explain why there were no differences in reaching asymmetry with aging and a trend for decreased error of the right arm (Fig. 3A). Conversely, our results suggest that when the lateralized specializations of each arm are performed, the compensatory activity may interfere with task performance, given that absolute performance was reduced for reaching and stabilizing. Thus our results partially contradict the dedifferentiation view of the HAROLD model because lateralization of stabilizing increases with aging.

Limitations.

An important limitation is that we did not measure muscle compliance, which may be asymmetric, may differ based on strength differences related to dominance (Lanshammar and Ribom 2011), and may increase in asymmetry with aging (Lehnert et al. 2014; Schimidt et al. 2014). However, while muscle asymmetry with aging was demonstrated in the legs (Lehnert et al. 2014; Schimidt et al. 2014), it is not yet evident that this also occurs in the arms. Furthermore, while our results illustrated no increase in strength asymmetry with aging, potential asymmetric changes in muscle compliance may also be linked to hemispheric motor asymmetries.

In addition, without unilateral stabilizing task conditions, it is difficult to determine if increased asymmetry in older adults can be attributed predominantly to greater competition between reduced intrahemispheric resources during bihemispheric control. Future studies are needed to investigate stabilizing performance in unilateral tasks to compare with our current results.

To determine if age-related increases in asymmetry were related to muscle strength, we measured grip strength given that it has been shown as a valid measure to characterize overall muscle strength and impairment of the upper extremity in older adults (Bohannon 1998; Bohannon and Andrews 2000). However, it is possible that muscle strength of the elbow and shoulder, as utilized during the present motor task, may have demonstrated differences in strength in the older adults included in this study.

Conclusions.

This study provides evidence for preserved asymmetry for reaching and greater asymmetry for stabilizing of interlimb differences in bilateral coordination with aging. This is the first report of preserved and increased task-related lateralization with aging within the context of a complementary bilateral task.

GRANTS

This research was supported by the American College of Sport’s Medicine Doctoral Student Research Award (to E. J. Woytowicz) and National Institute of Child Health and Human Development Grant R01HD059783 (to R. L. Sainburg).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

E.J.W., R.L.S., K.P.W., and J.W. conceived and designed research; E.J.W. performed experiments; E.J.W. analyzed data; E.J.W. interpreted results of experiments; E.J.W. prepared figures; E.J.W. drafted manuscript; E.J.W., R.L.S., K.P.W., and J.W. edited and revised manuscript; E.J.W., R.L.S., K.P.W., and J.W. approved final version of manuscript.

APPENDIX: EQUATIONS OF MOTION

Elbow interaction torque:

$$m_e{r}_e\sin (\theta +\varphi )\ddot{x}-m_e{r}_e\cos (\theta +\varphi )\ddot{y}-\left(l_s{m}_e{r}_e\cos \varphi +m_e{r}_e{}^2+I_e\right)\ddot{\theta }-(l_s{m}_e{r}_e\sin \varphi ){\dot{\theta }}^2$$

Elbow net torque:

$$\left(m_e{r}_e{}^2+I_e\right)\ddot{\varphi }$$

Elbow spring torque:

$$-\left[l_e\sin \left(\theta +\varphi \right){\text{F}}_x\right]+\left[l_e\cos \left(\theta +\varphi \right){\text{F}}_y\right]$$

Elbow muscle torque:

$$Elbow\hspace{0.17em}Net\hspace{0.17em}Torque–\left(Elbow\hspace{0.17em}Interaction\hspace{0.17em}Torque+Elbow\hspace{0.17em}Spring\hspace{0.17em}Torque\right)$$

Shoulder interaction torque:

$$\left\{(m_s{r}_s\sin \theta +m_e{l}_s\sin \theta )\ddot{x}-(m_s{r}_s\cos \theta +m_e{l}_s\cos \theta )\ddot{y}-\left[m_e{r}_e\left(l_e\cos \varphi \ddot{\varphi }+l_s\sin \varphi {\dot{\varphi }}^2+2l_s\sin \varphi \dot{\varphi }\dot{\theta }+l_s\sin \varphi {\dot{\theta }}^2\right)\right]\right\}$$

Shoulder net torque:

$$(I_s+m_s{r}_s^2+m_e{l}_s^2+m_e{l}_s{r}_e\cos \varphi )\ddot{\theta }$$

Shoulder spring torque:

$$-\left[l_s\sin \theta +l_e\sin \left(\theta +\varphi \right){\text{F}}_x\right]+\left[l_s\cos \theta +l_e\cos \left(\theta +\varphi \right){\text{F}}_y\right]$$

Shoulder muscle torque:

$$Shoulder\hspace{0.17em}Net\hspace{0.17em}Torque–Shoulder\hspace{0.17em}Interaction\hspace{0.17em}Torque+Shoulder\hspace{0.17em}Spring\hspace{0.17em}Torque+Elbow\hspace{0.17em}Muscle\hspace{0.17em}Torque$$

In the above equations, m_s and m_e are mass of upper arm and forearm, r_s and r_e are distance from the proximal joint to center of mass of upper arm and forearm, l_s and l_e are length of upper arm and forearm, and I_s and I_e are moments of inertia at center of mass of upper arm and forearm, where the subscripts s and e indicate upper arm segment and lower arm segment (including support and air sled device), respectively. θ is shoulder angle, ϕ is elbow angle, F_y is the component of spring force along the y-axis (medial-lateral), and F_x is the component of spring force along the x-axis (anterior-posterior).