The discharge of single motor units (n = 34) in the first dorsal interosseus muscle and the fluctuations in force during steady contractions were measured across a range of index finger abduction forces in old adults (77.1 ± 6.9 yr, n = 20). These results were compared with previously reported data on 38 motor units from young adults (25.7 ± 5.7 yr). Both minimal and peak discharge rates increased with recruitment threshold, but the strength of these relations was notably weaker for the old adults. Minimal discharge rates were similar for young and old adults (P = 0.77), whereas peak discharge rates were lower for old adults (P < 0.01). Consequently, the range of rate coding for each motor unit was substantially less for the old adults (7.1 pps) compared with the young adults (12.1 pps, P < 0.01). However, the variability in motor-unit discharge was similar for young and old adults; the coefficient of variation of the interspike intervals was similar at recruitment (old: 25.4%, young: 27.1%, P = 0.39) and declined with an increase in discharge rate (old: 13.2%, young: 14.2%, P = 0.21). Furthermore, the fluctuations in force during steady isometric contractions (2–95% of maximal force) were similar for young and old adults, except that the relative variability at the lowest force was greater for the old adults. A computational model of motor-unit recruitment and rate coding incorporated the experimental observations and was able to match the measured and simulated values for force steadiness across the operating range of the muscle.
As an explanation for the decline in motor performance that accompanies advancing age, a number of studies focused on the motor output from the spinal cord by recording motor-unit discharge (Christie and Kamen 2006; Erim et al. 1999; Galganski et al. 1993; Kamen and Roy 2000; Kamen et al. 1995; Laidlaw et al. 2000; Semmler et al. 2000; Soderberg et al. 1991; Tracy et al. 2005; Vaillancourt et al. 2003) and by assessing the intrinsic properties of motor neurons (Engelhardt et al. 1989; Morales et al. 1987). The results indicated that the change in some performance capabilities, such as the maximal strength of a muscle, is not influenced by changes in motor-unit activity. For example, although peak discharge rates are less during maximal contractions in old adults (Kamen et al. 1995), contractile speed is slower and this enables greater twitch fusion at lesser rates (Connelly et al. 1999) and old adults do not exhibit a conspicuous deficit in voluntary activation (De Serres and Enoka 1998; Klass et al. 2005). Rather, the decrease in the peak force that can be achieved by old adults during a maximal isometric contraction is largely attributable to the loss of muscle mass (Frontera et al. 2000a,b; Trappe et al. 2003).
However, the motor-unit pool does experience considerable remodeling with advancing age and this can result in fewer, but larger, functioning motor units in a muscle (Campbell et al. 1973; Gutmann and Hanzlikova 1976; Nikolic et al. 2001; Tomlinson and Irving 1977). One strategy to explore the functional significance of motor-unit remodeling has been to compare fine motor skills in young and old adults. Accordingly, it was shown that the ability of old adults to maintain a steady, submaximal force is impaired, especially at low forces (Galganski et al. 1993), and the deficit appears to be associated with elevated levels of discharge-rate variability (Laidlaw et al. 2000). Studies that examined the variability in motor-unit discharge rate, however, yielded mixed findings; some reported greater discharge-rate variability for old adults (Kornatz et al. 2005; Laidlaw et al. 2000; Tracy et al. 2005), whereas others found similar discharge-rate variability for young and old adults (Semmler et al. 2000; Vaillancourt et al. 2003). Furthermore, only in some experiments was reduced force steadiness associated with elevated levels of discharge-rate variability (Kornatz et al. 2005; Laidlaw et al. 2000; Tracy et al. 2005).
One of the difficulties in motor-unit studies is that the activity of only a few motor units can be recorded in each experiment. As a consequence, the implicit assumption in these studies is that the behavior of a few motor units adequately represents that of the entire motor-unit population. The mixed results on the contribution of discharge-rate variability to force steadiness thus might be caused by differences in the motor units that were used to represent the activity of the population. To assess the association between discharge-rate variability and force steadiness, Moritz et al. (2005) imported experimental observations on the discharge of single motor units into a computational model of motor-unit recruitment and rate coding (Fuglevand et al. 1993). The results indicated that discharge-rate variability changes across the activation range of a motor unit and that when this characteristic was specified in the model there was a strong association between discharge-rate variability and the steadiness of muscle force.
The combined approach of measuring then incorporating physiologically accurate behavior for individual motor units in a computer model of the entire motor-unit pool is the most effective practical means of evaluating the influence of motor-unit discharge characteristics on the output of the motor-unit pool (Jones et al. 2005; Moritz et al. 2005). Because Moritz et al. (2005) measured motor-unit activity only in young adults, the question remained as to whether or not discharge-rate variability contributes to the difference in force steadiness between young and old adults. The aims of this study were 1) to record motor-unit discharge characteristics across a wide range of forces in old adults and 2) to use the experimental measures in a model of motor-unit recruitment and rate coding to examine the association between discharge-rate variability and force steadiness.
The experimental protocols replicated the previous study with young adults (Moritz et al. 2005). In two separate experimental protocols, single motor-unit discharge and fluctuations in force during steady contractions were measured over a wide range of forces. Single motor-unit data were obtained for 34 motor units from 20 subjects (10 men and 10 women) who had a mean age of 77.1 ± 6.9 yr (range: 66–91 yr). The fluctuations in force were measured for 22 subjects (12 men, 10 women; 78.5 ± 7.3 yr) who also participated in the single motor-unit experiment. These data were contrasted with an expanded sample of data from the 22 young adults in the previous study (12 men 10 women; 25.7 ± 5.9 yr). All subjects were right handed, as verified by the Edinburgh Handedness Inventory (Oldfield 1971), and reported no known neurological disorders. The Human Subjects Committee at the University of Colorado approved the protocol and informed consent was obtained before participation from each subject.
Subjects were seated with the left shoulder abducted by about 0.79 rad (45°), and the forearm was restrained in a neutral position and rested on a platform. The elbow joint and forearm were immobilized with a vacuum pillow (Tumble Forms, Trenton, Ontario, Canada) and Velcro straps. The left hand was supported with the palm vertical, the third through fifth digits were flexed slightly at the metacarpophalangeal joints and restrained in a brace, and the thumb was extended vertically and held with a separate brace in the same plane as the palm of the hand at an angle of about 1.1 rad (60°) to the index finger. The index finger was secured to a hinged splint to maintain both interphalangeal joints extended and to constrain finger excursion to the abduction–adduction plane. The index finger was flexed at the metacarpophalangeal joint by about 0.1 rad (5°) to maximize the contribution of the first dorsal interosseus muscle to the abduction torque. The abduction force exerted by the index finger was measured with a force transducer aligned with the proximal interphalangeal joint. Low (0.049 V/N) and high (0.472 V/N) sensitivity transducers (model 13, Sensotec, Columbus, OH) were used to enable the measurement of forces from <2% maximal voluntary contraction (MVC) to maximal levels with a sufficiently high signal-to-noise ratio. Force was digitized with a Power 1401 (CED, Cambridge, UK) at 200 samples/s during the motor-unit experiment and at 1,000 samples/s during the force steadiness task, and then stored on a computer. A 17-in. computer monitor located at eye level in front of the subjects at a distance of 1.6 m was used to provide visual feedback of the abduction force exerted by the index finger.
Single motor-unit potentials were recorded from the first dorsal interosseus muscle using Formvar-insulated, stainless-steel wires (diameter: 50 μm; California Fine Wire, Grover Beach, CA) that were glued together at the recording tip with medical-grade cyanoacrylate and inserted into the muscle belly using a 27-gauge, 1.125-cm hypodermic needle. The needle was removed after the wires were inserted. The insulation was absent from only the recording tip of each wire and three wires were included in each recording electrode to permit alternative bipolar configurations. The quality of the single motor-unit recordings was optimized by using different pairs of recording wires and by making slight adjustments in the location of the electrode. The EMG signal was amplified ×5,000 or ×10,000, and band-pass filtered between 300 Hz and 8.5 kHz (S-series, Coulbourn Instruments, Allentown, PA). The motor-unit signal was sampled at 20K samples/s with a Power 1401 (CED) and stored on a computer. Single motor-unit potentials were identified on-line using both a dual-window discriminator (S-series, Coulbourn Instruments) and Spike2 software (CED). A reference surface electrode for the single motor-unit recordings was placed over the styloid process of the ulna (silver–silver chloride, 4-mm diameter).
EMG recordings of the antagonist muscle, second palmar interosseus, were also made to monitor agonist–antagonist coactivation during the single motor-unit experiment. Two Formvar-insulated stainless-steel wires (diameter: 50 μm; California Fine Wire) were inserted through the dorsal aspect of the hand into the belly of the muscle using a 30-gauge, 2.25-cm hypodermic needle. The back of the hand was cooled with an ice pack before insertion and the needle was removed after the wires were inserted. A reference surface electrode for the second palmar interosseus recordings was placed over the styloid process of the radius (silver–silver chloride, 4-mm diameter). The EMG signal was amplified ×1,000–5,000, band-pass filtered between 13 Hz and 1 kHz (S-series, Coulbourn Instruments) and sampled at 2,000 samples/s. Selective recordings from second palmar interosseus were assumed when the EMG signal was negligible during both a strong contraction with the thumb adductors (adductor pollicis) and a weak contraction with the middle finger abductors (second dorsal interosseus).
Subjects began both experimental protocols by performing several trials of the MVC task to determine the maximal force capacity of the first dorsal interosseus muscle. Participants were provided with verbal encouragement as they increased force from zero to maximum over a 3-s period and then held the maximal force for 1–2 s. Visual feedback of the abduction force exerted by the index finger was provided on the computer monitor and the subject's hand was closely observed by one of the experimenters to ensure that the task was performed correctly. The peak value from three or four trials was taken as the MVC force, provided it was within 5% of the peak value for another trial. A maximal contraction in the direction of index finger adduction was also performed for the purposes of normalizing the antagonist EMG.
The goal of the motor-unit experiment was to identify the potentials of a single motor unit that could be discriminated during brief contractions at multiple forces. Subjects were asked to perform three tasks during this experiment: 1) a graded minimal-rate task with audio feedback of discharge rate, 2) a ramp contraction in which force was increased gradually and continuously, and 3) a discrete isometric force task at many target forces.
The experiment began by the subject gradually increasing contraction intensity as the investigators observed the intramuscular EMG signal for the appearance of a candidate motor unit. The subject was provided with visual feedback of index finger force and audio feedback of the discharge for the motor unit being tracked. Once a unit had been identified, the subject was instructed to increase the force gradually until the motor unit became active and then to reduce the force slowly to identify the minimal rate at which the unit could discharge action potentials repetitively. This task, referred to as the graded minimal-rate task, was repeated three times, with each trial lasting between 5 and 20 s.
The recruitment threshold of the motor unit was characterized as the force at which the unit began to discharge action potentials repetitively during a ramp increase in index finger force. The target force for the ramp contractions was set at twice the force associated with the graded minimal-rate task. The ramp contractions were performed two to three times with the ramp lasting between 3 and 6 s for the up and down phases of the contraction.
Subsequently, an additional ramp contraction was performed to determine the peak force at which the discharge of the motor unit could be discriminated. Once the upper limit was identified, a series of about 10 target forces was determined across the range that the unit could be discriminated. The actual number of target forces varied, however, as a result of both the variable number of trials needed to identify the minimal discharge rate and changes that occurred in the recording conditions during the experiment. Successive target forces were presented on the visual display and participants were instructed to exert an index finger force to match the target and to do so without either exceeding the target force or generating a rapid contraction. Subjects were required to maintain the target force for 2–10 s, with briefer durations for the high forces. There was a rest period of ≥30 s between trials. The force targets were presented in an ascending order, except that the target forces were adjusted up and down in small increments (<1% MVC force) around the recruitment threshold to identify the force associated with the minimal discharge rate. The minimal rate obtained with this protocol is referred to as the discrete minimal discharge rate. After the series of contractions to the target forces, MVCs were performed to verify that the observed discharge rates were minimally influenced by fatigue.
FORCE STEADINESS TASK.
Subjects performed isometric contractions with the first dorsal interosseus muscle so that the index finger exerted an abduction force to match a series of target forces. Subjects practiced the task a few times at a moderate intensity before beginning the series. The target forces were presented in random order with two attempts at each of 2, 5, 15, 30, 50, 70, 85, and 95% of MVC force. Subjects were instructed to increase the abduction force to match the target indicated on the visual display and then to hold that force as steadily as possible for about 6 s. Visual feedback of target force and the index finger force was provided for the first 3 s. The entire force trace was shown to the participant after the completion of each trial. The gain of the force display was adjusted so that the target-force line was always at the same position on the monitor relative to zero force. Although the variation in gain likely influenced the subjects' force variability, only the nonvisual feedback data from each trial were included in the analysis. A minimal rest interval of 30 s was provided between each trial, with considerably longer rest periods after high-force contractions.
A model of motor-unit recruitment and rate coding, originally developed by Fuglevand et al. (1993), was used to simulate the isometric force produced by a pool of motor units with characteristics resembling the first dorsal interosseus muscle (Moritz et al. 2005; Taylor et al. 2002, 2003; Yao et al. 2000). The model was implemented in Simulink 6.4 for MATLAB (version 7.2, The MathWorks, Natick, MA). A detailed description of the simulation was previously published (Fuglevand et al. 1993; Taylor et al. 2002, 2003).
In brief, the model consisted of a pool of motor units with systematic variation in recruitment threshold, minimal and maximal discharge rates, twitch force, and twitch contraction time. Consistent with the most recent implementation of the model (Moritz et al. 2005), the simulated pool had 180 motor units (Jenny and Inukai 1983) and a recruitment range to an upper limit of 60% MVC. Variability in interspike intervals (ISIs) was included in the model by independently adjusting the timing of each discharge according to a normal distribution, with a specified coefficient of variation (CV) about the mean discharge rate. Motor-unit recruitment and discharge rate were determined by an excitation function that acted on all the units in the pool. Motor unit 1 was the first recruited, had the smallest twitch force [1 arbitrary unit (au)], and the longest twitch time (90 ms). In contrast, motor unit 180 was the last recruited, had the largest twitch force (100 au), and the briefest contraction time (30 ms). Each motor unit generated a twitch force in response to a single discharge and a tetanic force when the activation involved multiple discharges. The amplitude of each tetanus was defined by a gain function that depended on discharge rate. The simulated muscle force was calculated as the sum of all the motor-unit forces. Muscle force at each level of excitation was normalized to the force produced when all motor units were recruited and discharging at maximal rates (analogous to the MVC).
In early versions of the model (Fuglevand et al. 1993; Taylor et al. 2002, 2003; Yao et al. 2000), the minimal discharge rate was set at 8 pulses per second (pps) for all motor units, and the maximal discharge rate decreased from 35 pps for the first recruited unit to 25 pps for the last recruited unit. In the Moritz et al. (2005) version of the model, minimal discharge rate was increased linearly with recruitment threshold force from 7.6 to 17.9 pps and maximal discharge rate was also increased linearly with recruitment threshold force from 17.6 to 34.8 pps. Both changes to the model were based on experimental data. A further revision of the model, prompted by experimental observations, was to alter the variability in the ISI from a constant value for the CV of 20%. Variability in ISIs was set to a CV of roughly 30% at recruitment, which rapidly declined to 10% as the discharge rate increased.
In the Moritz et al. version of the model, minimal and maximal discharge rates were varied explicitly on the basis of the recruitment threshold force of a given motor unit. Similarly, the CV for the ISI was varied relative to the force above recruitment threshold force. This approach allowed rapid assessment of the influence of these parameters on the behavior of the model. It was subsequently suggested by Jones et al. (2005), however, that a more appropriate approach would be to vary these parameters in terms of the model excitation parameter (E), rather than in terms of the model output parameter, force. Accordingly, the minimal and maximal discharge rates and the CV for the ISI were specified in terms of the recruitment threshold excitation (RTE) parameter in the model. Maximal discharge rate was increased linearly with RTE using the original expression (Eq. 5) from Fuglevand et al. (1993). This equation was also used to vary minimal discharge rate (1) where i denotes the ith motor unit, n indicates the number of motor units in the pool, and MFRD represents the rate of increase in minimal discharge rate. A similar range of minimal (7.1 to 16.1 pps) and maximal (17.3 to 31.4 pps) discharge rates to Moritz et al. (2005) was achieved by using the input parameters for minimal firing rate (MFR) and peak firing rate (PFR) reported in the online material.1
Equations 2, 3, and 4 were devised to vary the CV of the ISI relative to E and RTE. These equations, with the input parameters reported in the on-line material, achieved the same rapid decline in CV for the ISI that was implemented by Moritz et al. (2005) (2) (3) (4)
The current version of the model included ISI distributions that were skewed rather than being symmetrically distributed about the mean (Calvin and Stevens 1968; Enoka et al. 1989; Matthews 1996). ISI distributions were drawn from a combination of normal and lognormal distributions, with the relative contribution of the lognormal distribution declining as Ein increased above RTE for each motor unit. Consequently, the ISI histograms were most skewed for newly recruited motor units when the CV for the ISI was high and became progressively less skewed as the discharge rate increased above recruitment levels, until the ISI histogram reached an entirely normal distribution. Consistent with the approach used by Fuglevand et al. (1993), a random selection of Z-scores was generated that were used to adjust the duration of each ISI along with a specified CV. Z-scores were randomly drawn from a combination of either normal or lognormal distributions according to a weighting function shown as Eq. 5. So as not to alter the specified mean discharge rate, the distribution of Z-scores was adjusted to have a mean of zero. A similar approach was used by Jones et al. (2002) to impose a Rayleigh distribution for ISIs (5)
A final version of the model was developed with parameters adjusted to simulate the motor-unit pool of an old adult (Table 1). The number of motor units was reduced to reflect the loss of motor units that occurs in old age (Brown et al. 1988; Doherty and Brown 1993; Larsson and Ansved 1995; Tomlinson and Irving 1977). The range of twitch torques was reduced, the average twitch torque was increased (Doherty and Brown 1997; Galganski et al. 1993; Kadhiresan et al. 1996; Larsson and Ansved 1995; McComas 1995; Spiegel et al. 1996), and twitch contraction times were increased (Andersen 2003; Doherty and Brown 1997; Kadhiresan et al. 1996; Klitgaard et al. 1990; Larsson et al. 1979; Lexell 1995; Lexell et al. 1988; Sugiura and Kanda 2004). Maximal discharge rates were also lowered (Kamen et al. 1995). The model did not include motor-unit synchronization because it was previously shown that it does not change with age in the first dorsal interosseus muscle (Kamen and Roy 2000; Semmler et al. 2000). All simulations were run 20 times so that the variability assigned to each parameter could influence the variability in the simulated forces.
Discrimination of single motor-unit potentials was verified off-line by visual inspection of each potential and by using the template-matching features of the Spike2 software (version 5.02, CED). ISIs >250 ms (<4 pps; n = 89, 0.23% of discharges) or <10 ms (>100 pps; n = 15, 0.04% of discharges) were excluded from the calculations of discharge rate. Long ISIs (<4 pps) likely arose from the brief cessation of motor-unit discharge, whereas very short intervals (>100 pps) exceed the rates normally observed during these types of contractions for human motor units (Bigland and Lippold 1954; De Luca et al. 1982; Kanosue et al. 1979; Tanji and Kato 1973) and likely resulted from discrimination error or double discharges. To determine the region over which to calculate the mean and CV for discharge rate during each contraction, the force plateau was identified as beginning when the force was within 90% of the target force and ending 1 s before the force dropped to <90% of the target force. The force and discharge-rate measurements were made over the intervening interval.
The recruitment threshold of each motor unit was determined by moving a 0.5-s window forward in time in 1-ms steps during the ramp task until the CV of the ISIs separating the potentials within the window was <50%. The force corresponding to the first discharge in this window was taken as the recruitment threshold of the motor unit.
The minimal discharge rate of each motor unit was measured in two tasks: 1) a graded decrease in force until a minimal rate was achieved and 2) discrete increments in target force around the recruitment threshold of the motor unit. The minimal discharge rate during the graded test corresponded to the lowest rate measured during a 2-s interval when the CV of ISIs was <50%. The minimal rate during the discrete test was identified by moving a 2-s window forward in 1-ms steps and noting the rate when both the CV for discharge rate was <50% and at least one discharge occurred both before and after the 2-s window. Data for the graded decrease in force with audio feedback were obtained for only 28 of the 33 total motor units because these units were recorded secondarily to a primary unit with a similar amplitude and made it difficult to provide accurate audio feedback to the subject.
Different types of distributions were fit to the ISI histograms for each motor unit for each discrete target force performed by the young and old adults. ISI histograms were constructed with a 1-ms bin width for a period of 1 to 250 ms. Normal, lognormal, and gamma distributions were fit to each histogram by maximum-likelihood estimation (Statistics Toolbox, MATLAB). The goodness-of-fit of the different distributions was evaluated by calculating the root-mean-square error (RMSE) between the fitted distribution and the ISI count for each bin. The RMSE was calculated for the entire ISI histogram and also the tail regions of the histogram lying outside of ±1 SD from the mean. Frequencies were collated for the distribution type that best fit each individual ISI histogram.
FORCE STEADINESS TASK.
The abduction force exerted by the index finger at each target level during the steady contractions was quantified for a 1-s period commencing 500 ms after visual feedback of the force was removed. The 1-s epoch was linearly detrended before the assessment of the CV for force to minimize the contribution of gradual drifts away from the target force. The mean force was calculated over this 1-s region for the two trials at each target level. For some trials, it was necessary to select a slightly later or earlier time window if the mean force was not sufficiently close to the target. The CV for force for the trial with the mean force closer to the target force was used in the analysis.
Simulations were run for eight target forces with 10- and 5-s contractions at each target force. Force data were measured during the middle 1 s of the 5-s simulated contraction and discharge-rate statistics were extracted for the middle 8 s of the 10-s simulated contraction.
Repeated-measures ANOVA and paired samples t-tests were used to compare the recruitment forces and discharge rates obtained during the ramp, graded, and discrete tasks. The relations between recruitment threshold and the minimal and peak discharge rates observed during the brief contractions at different target forces were characterized with linear regression analyses. Independent samples t-tests were used to compare the young and old adults. For some variables, ANCOVA assessed age-related differences to verify that detected differences in discharge-rate variables did not arise from sampling differences. Lilliefors test evaluated the goodness-of-fit of a normal distribution to the ISI histograms for each target force of each motor unit. The proportion of histograms that differed from a normal distribution was calculated and an arcsine-root transform was applied for the young and old data comparison of this index.
A two-factor, repeated-measures ANOVA compared the simulated and experimental measures of the CV for force (between-subject factor) at each of the eight target forces (repeated-measures factor). Post hoc analyses with paired-samples t-tests were used to identify the target forces that differed when interactions were identified by ANOVA. Regressions were also performed on the average CV for force at each target level as a means of quantifying the similarity between force output from the model and experimental data. All statistical procedures were performed with SPSS version 14.0 and SigmaPlot version 8.02 (SPSS, Chicago, IL). α was set at 0.05 and all reported values are means and SDs or 95% confidence intervals (CIs). Confidence intervals were calculated using a t-distribution because the sample sizes were <30.
Recordings were made from 34 motor units at 9 ± 3 target forces during contractions that lasted 11.3 ± 2.8 s. The target forces ranged from recruitment threshold to an average of 17.9% maximal voluntary contraction (MVC) force above recruitment (4.6–42.0% MVC). There were 117 ± 33 (range 68–198) discharges recorded at each force target over an average duration of 8.6 ± 2.4 s. EMG recordings during the experiments for 21 of the 34 motor units indicated that the activity of the antagonist muscle, second palmar interosseus, was minimal and consistent across the different target forces. The data were compared with that for 38 motor units from young adults from a previous study (Moritz et al. 2005) (Table 2). MVC force was significantly (P < 0.005) lower for the old adults (27.8 ± 8.0 N) compared with the young adults (35.1 ± 8.6 N).
Recruitment-threshold forces (Fig. 1A) measured during the discrete and graded minimal-rate tasks were similar to those for the ramp task (task main effect: P = 0.358). The discharge rates (Fig. 1B) recorded at recruitment did not differ for the ramp and discrete minimal rate tasks, but were significantly lower for the graded minimal rate task (task main effect: P < 0.001).
Motor-unit discharge rate increased with index finger abduction force for all units (Fig. 2A). Minimal discharge rates were similar for young and old adults (P = 0.773), whereas peak discharge rates were lower (P = 0.001) for old adults. Consequently, the range of rate coding for each motor unit of the old adults (7.1 ± 3.4 pps) was less than that observed for the young adults (12.1 ± 5.6 pps, P < 0.001) (Fig. 2B). Although the force range across which motor units were tracked did not differ significantly between young and old adults (P = 0.202), there was a trend for the mean force range to be smaller for the old adults (Table 2) and for the average peak force over which the motor units of young and old adults were tracked to be less (P = 0.134). The significant reduction in rate coding for old adults persisted when ANCOVA was used with either force range (P < 0.001) or peak force (P = 0.002) as covariates. Although it appears from the two data sets in Fig. 2B that motor-unit recruitment may also be compressed for old adults, the recruitment threshold force for the young and old adult motor unit samples did not differ significantly (P = 0.231).
Minimal discharge rates ranged from 6.2 to 18.2 pps and peak discharge rates varied from 9.9 to 27.9 pps. Regression analyses revealed that both minimal (r2 = 0.28, P < 0.002) and peak (r2 = 0.26, P < 0.003) discharge rate increased as a function of recruitment threshold (Fig. 3). Similar relations were observed with young adults for the minimal (r2 = 0.55, P < 0.001) and peak discharge rates (r2 = 0.51, P < 0.001), but the strength of these relations was notably weaker in old adults. For the young adults, an even stronger relation between discharge rate at recruitment and recruitment-threshold force was obtained for the ramp task (r2 = 0.71, P < 0.001) than for the discrete force task, but this was not observed for the old adults (r2 = 0.16, P < 0.05).
Variability in motor-unit discharge was similar for the young and old adults. The coefficient of variation for the ISI (relative variability) was similar at recruitment (old: 25.4%, young: 27.1%, P = 0.389) and declined to almost the same level with an increase in discharge rate (old: 13.2%, young: 14.2%, P = 0.210). The rate of change in discharge-rate variability was also similar for young and older adults; the slope of the linear relation between SD of ISIs and mean ISI (Fig. 4A) was 0.42 for old adults and 0.38 for young adults (P = 0.319). The SD of the ISI (Fig. 4B), which is an index of the absolute variability in discharge rate, was similar for the young and old adults at recruitment (old: 25.10 ± 10.25 ms, young: 27.97 ± 10.97 ms, P = 0.255) and reached a similar minimal value (old: 8.07 ± 1.87 ms, young: 7.54 ± 1.88 ms, P = 0.240). Relative variability in discharge rate, measured either as the CV of the ISI or the CV for discharge rate, declined progressively with a decrease in mean ISI (Fig. 4, C, E, and G).
The equations presented in Stein et al. (2005) were used to calculate discharge-rate variability in units of pulses per second (pps). The SD of discharge rate was high at recruitment, declined to a minimum as the discharge rate increased, but subsequently increased with further elevations in discharge rate (Fig. 4, D and F). Similar SDs of discharge rate were observed for the old (0.83 ± 0.26 pps) and young adults (0.88 ± 0.28 pps) at recruitment (P = 0.255). The minima reached by the old and young adults, however, differed (old: 0.52 ± 0.16 pps, young: 0.61 ± 0.21 pps, P = 0.054) and the SD of discharge rate differed significantly between groups at peak force (old: 0.65 ± 0.20 pps, young: 0.84 ± 0.32 pps, P < 0.005). Peak discharge rates were greater for the young adults (old: 17.7 ± 4.2 pps, young: 22.4 ± 6.9 pps). The CV for discharge rate was similar for the young and old adults at recruitment (old: 7.93 ± 2.75%, young: 8.67 ± 2.83%, P = 0.265) and declined to almost the same level at peak force (old: 3.73 ± 0.88%, young: 3.77 ± 0.81%, P = 0.802) (Fig. 4, E and G).
The shape of the ISI histograms for every discrete force target was also examined. At low discharge rates, the ISI distributions were skewed with a greater proportion of extralong ISIs, but became normally distributed as discharge rate increased. Sample data for two concurrently recorded motor units that display the changing shape of the histograms are shown in Fig. 5. Of the 322 histograms for old adults and 375 for young adults, the proportion of these that differed from a normal distribution was similar for young (0.41) and old (0.39) adults (P = 0.719). Normal, lognormal, and gamma distributions were fit to each histogram and the goodness-of-fit was quantified by calculating the root-mean-square error for each histogram bin. The best fit for most histograms was a lognormal distribution and there was a remarkable similarity in these data for the young and old groups. This is evident in the proportion of the ISI histograms from all motor units and target forces that were best fit by lognormal (old: 0.66 ± 0.23, young: 0.62 ± 0.21), Gaussian (old: 0.21 ± 0.17, young: 0.22 ± 0.18), or gamma distributions (old: 0.13 ± 0.14, young: 0.16 ± 0.14).
Force variability and simulations
When the minimal and peak discharge rates and the variability in discharge rate were specified in terms of the model input parameters E and RTE, the outputs of the model were similar to those reported by Moritz et al. (2005) (Fig. 6, A and B). Similarly, the use of lognormal and normal distributions for the ISIs rather than only a normal distribution did not alter the variability in the simulated ISIs (Fig. 6, C and F). The only difference was that the combined distribution (normal + lognormal) produced more long ISIs at low discharge rates. This addition had only a minor impact on the simulated force steadiness, but it did typically elevate the CV for force at 2% MVC force, bringing it closer to the experimental data than previous versions of the model. Discharge-rate variability had a substantial influence on the match between simulated and experimental measures of force steadiness (Fig. 6, G and H). The match between the experimental and simulated data was greater with a changing CV for the ISI (r2 = 0.93) compared with a constant CV for the ISI (r2 = 0.69).
Force-steadiness data were collected from a group of 22 young and 22 old adults (Fig. 7, A and B) who differed in the abduction force exerted by the index finger during a maximal voluntary contraction with the first dorsal interosseus muscle (young: 31.6 ± 8.5 N, old: 27.0 ± 6.8 N, P < 0.05). A significant main effect arising from age was observed for the CV for force (P < 0.05) measured at eight different target forces (2, 5, 15, 30, 50, 70, 85, and 95% MVC). The greatest difference between the young and old adults was at the lowest force (2% MVC, P < 0.05). An index of absolute variability, the SD of force, did not quite reach statistical significance (P = 0.074). The model was adjusted to approximate the properties of the motor-unit pool for old adults and the simulated measures of force were compared with the experimental measurements. Force-steadiness data from the old adults was closely matched (r2 = 0.92 for CV of force) by the simulated data (Fig. 7C). Comparisons of force-steadiness data from simulations of the motor-unit pools of young and old adults (Fig. 7, E and F) showed subtle, yet significant (main effect: P < 0.001 for CV and SD of force), differences that were reasonably consistent with the differences between the experimental measurements for young and old adults (Fig. 7, A and B).
The discharge characteristics of motor units with a wide range of recruitment thresholds were recorded in a hand muscle of old adults as the muscle performed isometric contractions at many target forces. The experimental measurements indicated that the old adults exhibited a compressed range of rate coding compared with young adults, but there was no difference between the two groups in discharge-rate variability. When these data were combined with other known changes that accompany aging, a computational model of motor-unit recruitment and rate coding was able to achieve a close match between measured and simulated values for force steadiness across the operating range of the muscle for the old adults.
Recruitment threshold and discharge rate
Motor-unit recruitment thresholds were assessed with three different tasks: a ramp increase in force, discrete constant-force targets, and a graded minimal-discharge task with audio feedback of motor-unit discharge. Consistent with young adults (Moritz et al. 2005) and a previous comparison of young and old adults (Spiegel et al. 1996), motor-unit recruitment thresholds were characterized equally well with all tasks. The discharge rates at recruitment, however, were significantly lower for the graded minimal-discharge task than for the ramp task or discrete-force targets. A similar association was observed for the young adults. Late adaptation in motor-unit discharge rate could explain the observation of lower discharge rates during the longer-duration graded minimal-discharge task compared with the briefer ramp and discrete tasks (Kernell 1965; Kernell and Monster 1982; Sawczuk et al. 1995). Presumably, the lower discharge rate during the graded minimal-discharge task is not accompanied by lower force because of the delayed mechanical response of muscle to reductions in discharge rate (Macefield et al. 1996).
Both minimal and peak discharge rates increased with recruitment threshold, but the strength of these relations was notably weaker for old adults compared with young adults. The decline in these relations for the old adults is likely a consequence of the motor-unit remodeling that occurs with aging, likely arising from the change in innervation number that disrupts the associations between recruitment threshold, contractile speed of the innervated muscle fibers, and discharge rate. Erim et al. (1999) reported a similar effect of aging on the relation between minimal discharge rate and recruitment threshold force.
Although Erim et al. (1999) also found that the relation between peak discharge rate and recruitment is disturbed in old adults, the associations differed from those observed in the current study. The observation that peak discharge rate increases with recruitment threshold contrasts with results from other studies on hand muscles that report low-threshold motor units achieving higher discharge rates than later-recruited motor units (De Luca et al. 1982; Tanji and Kato 1973). The data in the current study are limited by technical constraints that prevented the recording of motor-unit discharge up to maximal force. Other studies that assessed maximal discharge rates with ramp contractions to a set force level (De Luca et al. 1982; Duchateau and Hainaut 1990) may have underestimated the range of rate coding for high-threshold motor units if the force at the peak of these ramps was not sufficient to elicit maximal discharge rates for later-recruited motor units. Regardless of the methodological issues that limit the measurement of maximal discharge rates, the observation of reduced rate coding in old adults is consistent with other studies (Erim et al. 1999; Kamen et al. 1995; Knight and Kamen 2007; Patten et al. 2001). Furthermore, as implemented in our model of old adults (Fig. 7, G and H) and reported by Erim et al. (1999), the difference in discharge rates of consecutively recruited motor units was less for old adults.
The absence of an age-related difference in discharge-rate variability is consistent with some previous investigations (Semmler et al. 2000; Vaillancourt et al. 2003), but not others (Laidlaw et al. 2000; Tracy et al. 2005). Variability in motor-unit discharge rate was previously attributed to synaptic noise and its interaction with the time course of the postspike afterhyperpolarization (AHP) of a motor neuron (Calvin and Stevens 1968; Matthews 1999, 1996). From the limited number of motor neuron investigations with cells from old animals, there is no evidence of age-related changes in the intrinsic motor neuron properties that control the AHP and influence the susceptibility to activation by synaptic noise (Engelhardt et al. 1989; Morales et al. 1987)—thus it is perhaps not surprising that no age-related difference in the variability of motor-unit discharge rate was observed with the various measures that were used.
Rather, the similarity in discharge-rate variability for the young and old adults provides converging evidence to support the previous description of the pattern of discharge-rate variability in motor units across their activation range (Moritz et al. 2005). The results indicate that relative variability is high at recruitment and it declines as discharge rate increases, whereas absolute variability, measured as the SD of the number of motor unit discharges occurring in any 1-s period, increases with an increase in discharge rate (Jones et al. 2005; Stein et al. 2005). However, there is an initial rapid decline in absolute variability at low discharge rates just above recruitment threshold, which may explain why elevated levels of motor-unit discharge-rate variability were reported for old adults in some studies (Kornatz et al. 2005; Laidlaw et al. 2000; Tracy et al. 2005) but not others (Semmler et al. 2000; Vaillancourt et al. 2003). Accordingly, the variability in discharge rate for a motor unit should not be determined from a measurement at a single force.
Updating the Fuglevand model
The skew that is apparent in ISI histograms at low levels of excitation was described in previous studies (Calvin and Stevens 1968; Enoka et al. 1989; Matthews 1996; Poliakov et al. 1995). The variable shape of the ISI distributions was included in the model because of the possibility that variability in force might be exacerbated by a greater frequency of longer-duration ISIs (Duchateau et al. 2006). The skewed feature of ISI histograms was implemented in the model by drawing random ISIs from a combination of normal and lognormal distributions. Inclusion of a lognormal distribution provided the long tail observed in ISI histograms. The predicted effect of more variable motor-unit forces was apparent in the model output. For the range of twitch durations and twitch forces in the simulated motor-unit pool, however, this postulated contributor to force variability had only a minor effect.
There is recent evidence that the incidence of brief ISIs (≤10 ms), referred to as double discharges, may be reduced in old adults (Christie and Kamen 2006). The analysis in the current study excluded any double discharges, but this amounted to only 0.04% of ISIs recorded during the discrete constant-force contractions. Although twice as many double discharges were recorded for the young adults, the extremely low incidence is indicative of the irrelevance of this motor-unit behavior to steady-force contractions examined in this experiment. Accordingly, double discharges were not included in the computer model.
Force variability and old adults
The experimental protocols used in the present study for the assessment of force steadiness involved brief contractions performed in the absence of visual feedback. Under these conditions, the difference between young and old adults in both relative and absolute indices of force steadiness was relatively minor, with a clear difference only at the lowest force level (Galganski et al. 1993). Had we examined longer-duration contractions (Laidlaw et al. 2000), presented variable force targets (Vaillancourt and Newell 2003), or provided visual feedback (Sosnoff and Newell 2006a,c), there may have been more substantial differences in force steadiness. By observing only a short-duration, constant-force contraction without visual feedback, the intent was to assess steadiness under conditions of a relatively constant central drive to the motor-neuron pool. This permitted the investigation of the contribution of alterations in the motor-unit pool to age-related declines in force steadiness.
The close match between experimental and simulated data for the old adults confirms that the key features of the motor-unit pool of old adults that contribute to the decline in force steadiness with old age have been identified. It is notable that there was a difference in the output of the young- and old-adult versions of the model, despite the two models including the same degree of discharge-rate variability for each individual motor unit. The old-adult model had reduced rate coding, fewer motor units, and different distribution of twitch forces, which resulted in lower discharge rates at a given force. As a consequence, there were higher levels of discharge-rate variability across the motor-unit pool of old adults at low forces, which likely contributed to some of the difference in force steadiness between the young and old adults. Nonetheless, the slightly reduced force steadiness for the old-adult model was apparent across the entire operating range, despite similar levels of variability in discharge rate. Although previous studies suggested that a reduction in the number of motor units and an increase in motor-unit twitch forces does not reduce force steadiness (Enoka et al. 2003; Keen et al. 1994), these effects may emerge only when distributed across the entire population with realistic patterns of motor-unit discharge-rate variability. Furthermore, the measured and simulated differences in strength between young and old adults (Sosnoff and Newell 2006b) were associated with relatively minor differences in force steadiness.
The relatively similar force-steadiness profiles for the experimental and model data for both the young and old adults indicate that age-related changes in the input to the motor-neuron pool may be responsible for the more substantial decrements in steadiness often reported for old adults (Vaillancourt and Newell 2003; Vaillancourt et al. 2003). For example, changes in spinal reflex pathways (Earles et al. 2001; Kido et al. 2004), a reduction in the number of corticospinal fibers (Eisen et al. 1996), or potential alterations in monoaminergic drive to the motor-neuron pool (Christou et al. 2004) may influence the ability of old adults to provide a constant input to the motor-neuron pool.
The discharge-rate variability hypothesis
The finding of similar discharge-rate variability for young and old adults may seem to contradict earlier reports from our laboratory (Kornatz et al. 2005; Laidlaw et al. 2000; Tracy et al. 2005). Indeed, one motivation for conducting the current study was the conflicting evidence from other experiments, such as those by Semmler et al. (2000) and Vaillancourt et al. (2003), which reported no difference in discharge-rate variability between young and old adults. The primary reason for the apparent disparity is the rapid change in discharge-rate variability as the activation of a motor unit increases (Moritz et al. 2005). Importantly, the study by Moritz et al. (2005) established the potent influence of discharge-rate variability on force steadiness. These findings encouraged our continued investigation of discharge-rate variability in old adults, but indicated that it could be assessed accurately only by measuring the magnitude of discharge-rate variability within the same motor unit over a range of contraction intensities. The current study demonstrates that discharge-rate variability also exerts a potent influence on force steadiness in old adults, but found that this mechanism contributes minimally to the difference in force steadiness between young and old adults.
In summary, old adults exhibited reduced rate coding of motor-unit discharge compared with that of young adults, but the two groups had similar levels of variability in motor-unit discharge. In agreement with a previous study (Moritz et al. 2005), discharge-rate variability exerted a critical influence on the steadiness of force, especially at low forces. The inclusion of these characteristics of motor-unit discharge in a computational model of motor-unit recruitment and rate coding provided an accurate simulation of the force produced by old adults.
This research was supported by National Institute on Aging Grant AG-09000 to R. M. Enoka.
↵1 The online version of this article contains supplemental data.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2007 by the American Physiological Society