- Email: [email protected]
Independence of motor unit recruitment and rate modulation during precision force control
Pergamon
PII:
Neuroscience Vol. 88, No. 2, pp. 643–653, 1999 Copyright 1998 IBRO. Published by Elsevier Science Ltd Printed in Great Britain. All rights reserved 0306–4522/99 $19.00+0.00 S0306-4522(98)00248-6
INDEPENDENCE OF MOTOR UNIT RECRUITMENT AND RATE MODULATION DURING PRECISION FORCE CONTROL G. KAMEN* and D. C. C. DU Department of Exercise Science, Totman Building, University of Massachusetts, Amherst, MA 01003, U.S.A. Abstract––The vertebrate motor system chiefly employs motor unit recruitment and rate coding to modulate muscle force output. In this paper, we studied how the recruitment of new motor units altered the firing rate of already-active motor units during precision force production in the first dorsal interosseous muscle. Six healthy adults performed linearly increasing isometric voluntary contractions while motor unit activity and force output were recorded. After motor unit discharges were identified, motor unit firing rates were calculated before and after the instances of new motor unit recruitment. Three procedures were applied to compute motor unit firing rate, including the mean of a fixed number of inter-spike intervals and the constant width weighted Hanning window filter method, as well as a modified boxcar technique. In contrast to previous reports, the analysis of the firing rates of over 200 motor units revealed that reduction of the active firing rates was not a common mechanism used to accommodate the twitch force produced by the recruitment of a new motor unit. Similarly, during de-recruitment there was no tendency for motor unit firing rates to increase immediately following the cessation of activity in other motor units. Considerable consistency in recruitment behavior was observed during repeated contractions. However, firing rates during repeated contractions demonstrated considerably more fluctuation. It is concluded that the neuromuscular system does not use short-term preferential motor unit disfacilitation to effect precise regulation of muscular force output. 1998 IBRO. Published by Elsevier Science Ltd. Key words: motoneuron, motor unit, recruitment, rate coding, firing rate.
Two principal mechanisms have been identified to account for the ability of the muscle to vary its force output. One mechanism involves the recruitment of new motor units according to increasing size—the ‘‘size principle’’.13–16 By increasing the number of active motor units within the motoneuron pool, the total force output increases. Aside from the recruitment of new motor units, the major mechanism by which muscular force is graded involves the modulation of motor unit discharge rates—rate coding.30 Considerable research has supported the size principle of motor unit recruitment, however, our knowledge of the factors that control discharge frequency is not nearly as developed. It is known that an approximately sigmoid relationship exists between firing frequency and force output.3,21 For long-term effects, motor unit discharge rate increases with increasing force output,2,30,31 although during relatively long contractions it decreases when constant force is maintained.33 However, the firing rate of human tibialis anterior muscle shows no progressive decline when *To whom correspondence should be addressed. Abbreviations: FDI, first dorsal interosseous; ISI, interspike interval; MVC, maximum voluntary isometric contraction.
the maximal voluntary contraction is sustained during acute deafferentation.27 Units from similar motoneuron pools tend to exhibit similar fluctuations in their firing rates.18 In addition, there are many short-term factors that affect the relationship between firing frequency and force output, such as membrane afterhyperpolarization23 and recurrent inhibition.17,24 However, the role that rate coding might play in compensating for these short-term influences on muscular force is not fully understood. Many studies have been conducted to investigate either recruitment or firing rate modulation in the motor control system, however, few studies have considered how firing rate modulation and motor unit recruitment interact to produce smooth muscular force. In human voluntary contractions, Broman et al.4 described an interesting phenomenon that suggested an alternative mechanism by which subtle changes in muscular force could be produced by the human motor system. Their experiment was conducted on the tibialis anterior and the first dorsal interosseous muscles. Subjects were asked to produce isometric contractions involving gradual increases in muscular force. When new motor units were recruited, the firing rates of the previously-active motor units were slowed. They suggested that the neuromuscular system was capable of disfacilitating
643
644
G. Kamen and D. C. C. Du
active motor units in order to lessen the incremental force contribution from an entire new motor unit and increase the smoothness of the total force output. However, the idea that individual motoneurons can be selectively disfacilitated is a bit disturbing, as it requires a considerable increase in control complexity. Indeed, a key feature of the size principle is its ability to present a mechanism by which the CNS is freed from the task of selecting a different population of motor units depending on force requirements. The present study was designed to further investigate the interaction between motor unit recruitment and rate coding during slowly-increasing isometric contractions requiring precision force production. If the goal is to produce a finely graded force output, is firing rate reduction a viable means of blending the contribution of additional twitch force from a newlyrecruited motor unit with the summated forces produced by the active units? We studied the influence of recruitment on firing rate modulation by tracking many active motor unit firing rates simultaneously during new unit recruitment. Using intramuscular recording of motor unit activities from the first dorsal interosseous muscle (FDI), we computed the firing rates immediately before and immediately after the recruitment of new motor units when the force output was smoothly increasing. These pre- and postrecruitment firing rates were then statistically compared to study possible inhibitory effects from the newly recruited motor unit to the firing rate of previously active motor units. We also investigated the interaction between motor unit de-recruitment and rate coding during slowly-decreasing isometric contractions during precision force production. The firing rates immediately before and immediately after the de-recruitment of a high threshold unit were calculated. The possible compensatory effects were then examined by comparing statistically the pre- and post- de-recruitment firing rates. EXPERIMENTAL PROCEDURES
Subjects Six adult subjects (20–24 years) consented to participate in the study. These individuals had no known history of neuromuscular disorders. All experimental procedures described were approved by the University Institutional Review Board. Procedures Subjects were seated comfortably at a table with their forearm and hand secured in a specialized apparatus designed to measure isometric force during index finger abduction. Maximum voluntary isometric contractions (MVCs) were measured while the subject was instructed to abduct the index finger at maximal effort. A visual trajectory was displayed on a computer monitor for two tasks. One task required the subject to slowly increase the isometric force at a rate of 4% MVC/s up to 30% MVC, then reduce the force at the same rate to resting level. The other task was similar, except that the range of force was between 10% and 40% MVC. The visual trajectory displayed on the
monitor preceeded the subject’s force by 0.25 s and progressed at the same speed as the subject’s trajectory. This real-time display facilitated subject’s force control to yield precise changes in muscular force. Each triangular trajectory was traced by the subject for four consecutive times during the same recording. A single trial then, consisted of four triangular trajectory contractions. Familiarization All subjects were given at least three days of practice to become familiar with the trajectories. Prior to motor unit recording on the fourth day, an evaluation was conducted to verify the proficiency of trajectory production. The experiment was conducted only when the subject could follow the desired trajectory and keep the computed mean square error within a 3% MVC range. Figure 1 shows a typical force recording and the prototype trajectory is superimposed. Force recordings During isometric contractions, the force signal was acquired using a strain gauge transducer (Interface, model MB-10). The force signal was d.c. amplified, filtered (d.c.=1 kHz), and stored on an instrumentation recorder (TEAC MR-30) for later digitization at 50 samples/s. Motor unit recordings The technique for recording and identifying motor unit activity has been described recently.19 Following the assessment of maximal force production, a four-wire needle electrode was inserted into the belly of the first dorsal interosseous, the index finger abductor. This needle electrode consists of a small diameter (25-gauge) stainless steel cannula containing four 50 µm platinum-iridium wires emerging from a side port 7.5 mm from the tip of the electrode. The wires in the side port were arranged in a square array with approximately 200 µm on each side. The signals obtained from the needle electrode were fed through three separate differential amplifiers (Dantec Counterpoint, input impedance: 200 MÙ, bias current: 25 pA), bandpass filtered (1–10 kHz; 3 dB), and recorded on the same instrumentation recorder (TEAC MR-30) for later digitization at 51,200 samples/s. We also manipulated the location of the needle electrode to sample different motor units in the muscle. For each subject, motor units were sampled from four to six different sites. There were usually three to six distinct motor units that could be detected in each contraction. After signal digitization, an epoch detection program algorithm separated the motor unit action potentials from the baseline signals. These three-channel motor unit signals were then processed by a semi-automatic microcomputerbased procedure which identified individual motor unit discharges based on their activity shapes in all three channels. After motor unit spikes were recognized, the firing information of each motor unit action potential train was extracted in order to obtain firing rate estimates. A sample of the motor unit firing activities obtained with the associated force output is illustrated in Fig. 2. Firing rate estimates In order to evaluate the influence of motor unit recruitment on the firing rate of active motor units, a firing rate estimate method is needed. Many reports have used unique, customized estimate procedures for motor unit firing rate extraction to address specific research questions.22,32,33 However, there is no consensus regarding the optimal procedure to estimate motor unit firing rate. Differences in the procedures used to calculate firing rate can lead to marked discrepancies in the estimated values of a unit’s firing rate. One recent firing rate estimate procedure used
Rate modulation and precision force control
645
Fig. 1. The prototype force trajectory of the experiment is superimposed with one typical force production during the study. The rate of force change was 4% MVC/s for both force increasing and decreasing ramps.
Fig. 2. Motor unit discharge history and the recorded force trace (solid, heavy line) in one segment of slow, increasing isometric contraction.
the reciprocal of the mean of five consecutive interspike intervals (ISIs).19 However, the summation of five ISIs in the human FDI at the 30–40% MVC values in this experiment would yield a window size of approximately 300 ms, and possibly as long as 800 ms. This time window might be too long to detect any short-term firing rate changes due to new motor unit recruitment. A fixed length weighted Hanning window such as the 800 ms duration window used by Broman et al.4 might also be too long to detect any short-term influence of new motor unit recruitment on
motor unit firing rates. Additionally, an 800 ms Hanning window can prolong the duration and consequently mask the appearance of short-term firing rate variations in the time domain. A similar problem is obtained with the use of moving average estimates of firing rate due to the fact that the moving average is a rectangular filter.41 We chose to use a modified boxcar algorithm, developed for cardiac heart rate variability analysis (from Berger et al.,1 see Appendix), for the short-term firing rate estimate. The selected boxcar technique allows for the accurate
646
G. Kamen and D. C. C. Du
Fig. 3. Motor unit recruitment threshold consistency in repeated contractions. The motor unit recruitment threshold in one contraction was plotted against that of the next contraction. The same motor unit was recruited at a similar force level regardless of its recruitment threshold.
estimate of firing rate during short durations comparable to one ISI. In order to make comparison with previous reports, the firing rates immediately before and immediately after the new motor unit recruitment were also calculated by applying both ISI averaging (the reciprocal of the mean of five ISIs) and Hanning window firing rate estimates, using a filter width of 800 ms. RESULTS
Consistency of motor unit recruitment thresholds The consistency of motor unit recruitment thresholds was assessed for all motor units available for analysis. The recruitment threshold was taken as the force score corresponding to the time of the first action potential. From repeated contractions, the recruitment thresholds of motor units observed during one contraction were compared with those of the next contraction whenever these same motor units were available. The scattergraph in Fig. 3 shows that the motor unit recruitment thresholds of one contraction were comparable to these same motor unit recruitment thresholds of the next contraction trial during the same trial. There was a linear correlation (r2 =0.90, n=226) between the recruitment threshold of one contraction and that of the same motor unit in the next contraction. We also inspected the motor unit recruitment thresholds during repeated contractions when the force production was varied in ranges of either 0% to 30% or 10% to 40% MVC. The statistical analysis of Student’s t-test revealed that there were no significant differences of the recruitment thresholds between contractions conducted in either the 0% to 30% or the 10% to 40% MVC force range (n=85; t=1.35; P=0.18). Motor unit discharge rates during new unit recruitment The data selected included 47 contractions in which 256 pairs of motor units (an active motor unit and a newly-recruited motor unit) were identified for
Fig. 4. Motor unit firing rates estimated by the ISI averaging method. (A) The scatter distribution of pre- and postrecruitment motor unit firing rates. (B) The relationship between firing rate difference at the new motor unit recruitment and the threshold difference between the active and newly-recruited motor units.
analysis. In these 256 cases, five consecutive ISIs of the active motor unit were available for analysis both before and after new motor unit recruitment. We also recorded 169 cases in which the motor unit was firing for at least 800 ms when a new motor unit was recruited, and maintained its firing for at least 800 ms after recruitment. These 169 sets of ISI data from 47 contractions were used for the weighted Hanning window estimate of firing rate (filter width=800 ms). Motor unit firing rates before and after new motor unit recruitment were computed using the ISI averaging (Fig. 4A) and the Hanning window filter method (Fig. 5A). As illustrated in these scattergrams, there was no consistent change in the firing rate of active motor units when a new motor unit was recruited. While some motor units did decrease firing rate when other motor units were recruited, there were also many units that exhibited an increase in firing rate. The differences between post- and prerecruitment firing rates are plotted as a function of the differences in recruitment threshold in Figs 4B and 5B using the ISI averaging estimation and Hanning window filter algorithm, respectively. These figures indicate that similar trends in the data were obtained irrespective of the method used to compute firing rate.
Rate modulation and precision force control
647
Fig. 5. Motor unit firing rates evaluated by the weighted Hanning window filter. (A) The relation between pre- and post-recruitment motor unit firing rates. (B) The motor unit firing rate changes against the corresponding threshold differences.
Fig. 6. Motor unit firing rate estimates derived by a modified boxcar algorithm. (A) The distribution of motor unit firing rates evaluated before and after recruitment. (B) The relationship between the firing rate changes and the recruitment threshold differences.
Statistical analysis of firing rate changes
Short-term discharge rate estimate using the boxcar algorithm
Using the reciprocal of the mean of five consecutive ISIs, the mean firing rate of the already-active motor units was 0.254.403 pulses/s (n=256) faster following the recruitment of a new motor unit, and a paired Student’s t-test revealed no significant difference in firing rate (t=0.89; P=0.37). From the Hanning window filtering calculations, the firing rate difference immediately following new recruitment was 1.965.262 pulses/s (n=169), and this increase in firing rate was statistically significant (t=4.85; P<0.05). Thus, rather than obtaining a decrease in firing rate, both of these analyses revealed an increase in firing rate, likely attributable to the requirements of the increased force demand over the relatively long time-course used in these two analyses. The observation that the Hanning window calculation was almost eight times greater than the reciprocal of the five ISIs (1.96 pps vs 0.25 pps) supports our concern that the 800 ms Hanning window uses a time-course that is too long to be sufficiently sensitive to shortterm disfacilitation of previously-active motor units. Firing rate estimates obtained with this Hanning window appear to be considerably more sensitive to longer-term increases in force production than short-term disfacilitation effects.
From the same 47 contractions, we were able to obtain 266 instances that a motor unit was firing at least 100 ms before a new motor unit was recruited, and the activity continued for at least 100 ms. We applied the modified boxcar procedure as described in the Appendix to these 266 motor units to compute the instantaneous firing rates before and after new motor unit recruitment using an interval of 100 ms. The motor units in the FDI muscle begin firing at roughly 10 pulses/s at recruitment.11,30,33,41 Motor unit firing rates in this muscle are normally in the range of eight to 40 pulses/s when the rate of force production is moderate and the maximal force level of the contraction does not exceed 50% MVC.19 Consequently, a short-term firing rate evaluation interval of 100 ms should be sufficient to encompass short duration ISI variations when firing rates fall into these ranges. If new motor unit recruitment causes only a few ISIs to be lengthened after the moment of recruitment, the firing rate estimate using the shorter 100-ms window will be a more powerful tool for differentiating this short-term variation of motor unit firing rate. The results of the boxcar analysis are shown in Fig. 6. The pre- and post-recruitment motor unit firing
648
G. Kamen and D. C. C. Du
Fig. 7. Active motor unit pre-recruitment firing rate in repeated contractions. Each motor unit’s firing rate estimated before a new motor unit was recruited was plotted against the same unit’s firing rate estimate before the same new motor unit was recruited again in the next contraction.
Fig. 8. Motor unit firing rate variations. In repeated contractions from the same subject, the change in firing rate for five different motor units are presented over the course of four consecutive contractions in two different force ranges.
rates were approximately evenly distributed along the line describing x=y. When the motor unit firing rates before and after new motor unit recruitment were compared by paired Student’s t-test, the results showed that there were no significant differences between the post- and pre-recruitment firing rates with the mean of the differences equal to 0.373.842 pulses/s (n=266; t=1.55; P=0.12).
ous ramp force productions in which force varied from 0% to 30% MVC. The sequences of H1 to H4 were drawn from the four consecutive force productions executed within the 10%–40% MVC range. There were five distinct motor units identified in each of the four ramp force contractions, and these same five units were also identified in the two different force ranges. The motor units were numbered according to the order of recruitment thresholds, i.e. the smaller motor unit numbers were given to the lower threshold units. When the same new motor unit was recruited, the firing rate changes varied considerably for all five motor units. At the first ramp force production of the 0%–30% MVC range, all five motor units increased their firing rates at new unit recruitment. However, in the last ramp force production of the 10%–40% MVC range, these five motor units decreased their firing rates when the same new unit was recruited again. In other ramp contractions, as can be seen from the figure, some motor units increased their firing rates while other motor unit firing rates decreased at the time of recruitment. In addition, the direction of firing rate change in repeated contractions varied considerably, regardless of the recruitment threshold and the range of force production. The sample illustrated in Fig. 8 was selected on the basis of the relatively large number of motor units (five) available for analysis from the same subject during these consecutive contractions.
Firing rate analysis of repeated contractions The repeatability of motor unit firing rate estimates obtained immediately prior to the recruitment of new motor units (‘‘recruitment rate’’34,35) was also assessed using the repeated contractions. The modified boxcar algorithm was used for the remainder of the firing rate estimates in the following analyses. During the repeated contractions, an active motor unit whose firing rate was estimated immediately prior to the recruitment of a new motor unit was examined in the next contraction at the precise point in the isometric force recording where the newlyactive unit was again recruited. These two motor unit firing rate estimates of the same unit in consecutive contractions are plotted in Fig. 7. Repeated firing rate estimates revealed a moderate correlation (r2 =0.54, n=226) with considerable dispersion. We also scrutinized the stability of active motor unit firing rate changes at the moment of new unit recruitment in repeated contractions. There was no observable consistency of motor unit firing rate change in repeated contractions. A sample of these firing rate changes in the consecutive contractions is shown in Fig. 8 where the firing rate changes are the differences between post- and pre-recruitment firing rates. Two different force ranges are illustrated, including one contraction conducted within 0% to 30% MVC and the other executed between 10% and 40% MVC. In each contraction, there were four sequential ramp force variations. In the figure, the series labeled L1 through L4 were the four continu-
De-recruitment analysis The relationship between motor unit recruitment and de-recruitment behavior was investigated in the same ramp contraction. The force values corresponding to the instances of the first and the last action potential were used as the recruitment and de-recruitment thresholds, respectively. The statistical analysis of the recruitment threshold and the de-recruitment threshold revealed a high Pearson correlation (r2 =0.81; n=263). The inter-relationship
Rate modulation and precision force control
649
Fig. 9. The scattergram of recruitment threshold versus de-recruitment threshold. There were significant correlations between recruitment and de-recruitment thresholds in one ramp contraction.
between recruitment and de-recruitment thresholds is illustrated in Fig. 9. To what extent might the firing rate of active units be altered when a high-threshold motor unit was de-recruited? We observed more than 200 pairs of motor units when the active motor units were still firing and maintained their activities for at least the interval required by the firing rate estimate method (boxcar algorithm, width=100 ms) at the highthreshold motor unit de-recruitment. Firing rate estimates were computed before and after the instances of motor unit de-recruitment. The results (see Fig. 10A) indicated that motor unit firing rates had a tendency to decrease continuously when a high-threshold unit was de-recruited: mean postde-recruitment vs pre-de-recruitment firing rate was 1.153.647 pulses/s (n=285). Differences between post- and pre-de-recruitment firing rate were compared by paired Student’s t-test which was statistically significant (t=5.32; P<0.05). One might suspect that the tendency to change firing rate prior to de-recruitment would be greatest among motor units with the greatest difference in recruitment thresholds. However, the results showed that there was no correlation between the change in firing rate and the relative change in recruitment threshold (see Fig. 10B). DISCUSSION
The results of this investigation refute the idea that the neuromuscular control system compensates for new motor unit recruitment by disfacilitating the active units in order to maintain precision during force production. Motor unit firing rates of alreadyactive motor units were estimated using several procedures, including the implementation of a boxcar algorithm (see Appendix) to ensure accuracy of short-term firing rate estimates. There are several reasons why the data presented obviate the need to postulate the existence of motoneuron disfacilitation while neighboring motoneurons are recruited.
Fig. 10. The firing rate variation at motor unit derecruitment. (A) The scattergram of motor unit firing rates estimated before and after de-recruitment. (B) The relationship between firing rate change and the difference of recruitment thresholds.
The problem of force regulation The CNS is assigned the task of smoothly grading force output by jointly controlling the numbers of active motor units and their respective firing rates. However, precision force tasks require the integration of individual twitch forces generated by all active motor units to produce a continuous, smooth contraction. So the problem faced by the CNS requires the translation of randomly-active digital units into a smooth analogue output. The force gradation problem is made more arduous when sufficient extra force is required to justify recruitment of an additional motor unit. At the moment of new motor unit recruitment, the additional twitch force has to be blended with the force production of the active units. Both the intensity of this new twitch force and the timing of the twitch with respect to the already-firing motor units have to be accommodated to result in a smooth, finelygraded force output. In this study, we investigated this aspect of the neuromuscular control mechanism, the influence of motor unit recruitment on active motor unit firing rates, when the muscle force output was slowly increased during a linear ramp trajectory. The influences of the rate of tension development The rate of isometric tension development influences motor unit recruitment as well as firing rate
650
G. Kamen and D. C. C. Du
modulation. The motor unit firing rate increases as the speed of the contraction increases.28 All motor units have lower recruitment thresholds with increases in the speed of force production.5,28 Although the increasing rate of isometric tension development lowers the recruitment threshold of motor units, the mechanical twitch tension from the newly recruited motor unit contributes at the same force level regardless of the speed of contraction.5 Whenever a new motor unit is recruited, the mechanism of force regulation adapts to the additional twitch tension in the same way throughout the whole range of force production. With this invariant recruitment framework, the control mechanism of the neuromuscular system can be simplified without addressing each additional motor unit twitch force differently. The force level at the onset of recruitment was used as the recruitment threshold. To minimize the effect of lowering motor unit recruitment threshold caused by the speed of contraction, we explored the minimal rate of force production that subjects could perform and 4% MVC/s was the slowest rate of force change not associated with moderate force variation. It is known that motor unit recruitment thresholds do not change when the contraction speed is sufficiently slow.10 Additionally, the maximal number of distinct motor units can be recruited successively only when a precision, isometric contraction is conducted at an extremely slow rate. However, only small numbers of motor units are recruited and they fire slowly at low force level. Due to the limited number of twitch tensions being fused, an increasing force variation is often observed at these low forces.34 This intrinsic property of motor unit twitch tension, one discrete twitch for each firing, might well define the minimal rate of force development. Are additional force regulatory mechanisms necessary? By contrasting the newly recruited unit’s tension with the summation of the smaller units’ force output, Henneman and Olson14 indicated that the twitch contraction from the newly-activated motor unit added tension at a constant proportion to the total cumulative force. Milner-Brown and Stein29 also observed this approximate constancy of relative recruitment steps in the first dorsal interosseous muscle. They observed that the twitch tension of the new motor unit varied nearly linearly as a function of the level of voluntary force at which the unit was recruited. Thus, it is possible that this constancy between force requirement and current force level obviates the need to theorize the existence of additional force regulatory mechanisms. Is compensatory disfacilitation an operative mechanism? If compensatory disfacilitation were an active control mechanism, there are several reasons it should
have been observed in the present study designed to reveal such a mechanism. First, small muscles rely more on firing rate modulation than larger muscles.25 Any phenomenon that produces momentary fluctuations in firing rate should be especially apparent in the small FDI muscle chosen for this investigation. Second, firing rate modulation becomes a more important mechanism in force graduation at intermediate force levels.30 The force levels varying up to 40% MVC should have optimized the use of compensatory disfacilitation during new recruitment. Third, compensatory disfacilitation should be especially attractive in a muscle with smaller numbers of motor units; the first dorsal interosseous muscle contains approximately 120 motor units.8 Small numbers of motor units would exacerbate the digital-to-analogue translation problem. Fourth, the task of increasing and decreasing force at a rate of 4% MVC/s requires considerable precision. This gradual increase in force and rate of motor unit recruitment should emphasize the rate coding mechanism and serve as an ideal environment for a mechanism such as compensatory disfacilitation. Finally, the boxcar method of computing discharge rate should offer the greatest sensitivity to identify any momentary rate changes. Temporary cessation of motor unit activity has been observed in some situations. For example, when a motor unit is recruited during a small but rapid increase in muscular force, it often stops firing when the force level reaches a new stable value and resumes continuous discharge only when the force output reaches a new tonic level.11 We do know that the increased rate of force production results in earlier motor unit recruitment with increasing firing rate.5,28 After a short force increase, it is possible that reduction of the supraspinal presynaptic excitation to the motoneuron pool produces a cessation of highthreshold unit firing and a reduction of active unit firing during the subsequent plateau contraction. Perhaps this impulsive disfacilitation was observed by Broman et al.4 when they asked subjects to hold the level of force constant immediately after recruitment. The lengthened motor unit ISIs could have resulted from the abrupt change in the rate of force production. Repeated contractions The results demonstrated that recruitment thresholds were stable in repeated contractions in which the rate of tension increase was kept constant. This is consistent with the results from other reports.10,34,38 Moreover, motor units were also recruited at similar levels of force production irrespective of the range of contraction. On the other hand, the active motor unit firing rates at the moment before a new motor unit was recruited were distributed in a wider range than the motor unit recruitment thresholds. It has been suggested that recruitment provides a coarse force gradation, while firing rate modulation complements
Rate modulation and precision force control
the control mechanism by providing a finer gradation of muscle contraction.34 Based on the invariant recruitment framework, the smallest motor unit is the first unit activated, the magnitude of the tension contribution and the moment when the twitch tension will be fused with the total force production can be predicted from the size of the unit. After motor unit recruitment, the rate coding mechanism controls the tension output of the unit. With a wide range of firing rate, the force contribution of the unit can be finely modulated.
De-recruitment In congruence with previous observations, a positive relationship (r2 =0.81) was observed between motor unit recruitment threshold and de-recruitment threshold.39 Motor unit firing rate estimates computed before and after de-recruitment yielded similar findings to the recruitment analysis: the cessation of motor unit activity does not alter the firing rate decrement of other motor units. The neuromuscular system controls the cessation of motor unit activity without compensatory control of active motor unit firing rates. The distribution of motor units in the muscle has been found to be highly skewed, having many smaller units and progressively smaller numbers of large units.14,29,37 Newly-activated motor units are approximately constant in size compared to the total tension of already-activated units. It has been suggested that this observation is in accordance with the Weber-Fechner relationship of sensory psychophysics,12,20,40 under which the relationship between the cumulative force (logarithmic) and the number of recruited motor units (linear) is logarithmic. The extension of the Weber-Fechner principle to motor unit recruitment is logical, as there are many smaller motor units recruited at low force levels and fewer units are recruited as the force level increases. Such a logarithmic relationship has indeed been illustrated in cat muscles.3,9 One way to achieve the finest gradation of force production is to maximize the number of motor units. However, this is not the organization used in the neuromuscular control system. Rather, motor units are recruited at different degrees of CNS excitation and the recruitment mechanism activates the smallest units first, ratcheting up to the largest units according to the size principle. This motor unit recruitment scheme, based on the Weber-Fechner principle of sensory psychophysics, was considered by Kernell20 as a possible means through which the neuromuscular system can achieve steady and finely gradable force production using recruitment and rate coding. The number of motor units is minimized so that smooth force gradation is achieved when a new twitch tension is added to the total force production according to the Weber-Fechner principle. Therefore,
651
the motor unit recruitment mechanism itself is furnished with the capability to generate smooth, graded force production. However, recruitment of all the motoneurons of the pool can only produce 10–25% of the maximal tension generated when all the units’ firing rates are at the minimum level to cause twitch tension fusion.10 At higher force levels, motor unit firing rate modulation contributes more to force production than recruitment.30 The neuromuscular system integrates the motor unit twitch tensions by enforcing predominant asynchronous motor unit firing behavior and asynchronous fusion of unit twitch tensions.26,36 Indeed, it has been suggested that one of the effects of Renshaw recurrent inhibition is to enhance the asynchronous firing between active motor units.42 Moreover, there are observations that synchronous activation of adjacent motor units increases the total force output non-linearly by force argumentation.6 The control mechanism of rate coding is far more complex than that of recruitment as illustrated in the results. Firing rates of the active units when new units are recruited varied somewhat in different contractions whereas the recruitment thresholds were consistent during repeated contractions. One possibility is that each motor unit has an inherent discharge variability, including a discharge variability at the onset of recruitment. Such variability could be due, in part, to the spatiotemporal structure of synaptic input to each motoneuron. Although the proportion of recruitment and rate coding is different for each individual muscle group, the demand of finely graded force production and the neuromuscular control mechanisms are similar. From the above discussions, whenever the motor unit recruitment is wellorganized, the condition of smooth force production can be achieved. Superimposed on motor unit recruitment, the rate coding mechanism regulates the firing time of active motor units. However, the manner in which rate coding is regulated so that each twitch is fused for smooth, finely graded force output will require further exploration. CONCLUSIONS
The results of this experiment provide no evidence for the idea that the motor control system compensates for new motor unit recruitment by disfacilitating the active units. Compensatory disfacilitation would be computationally expensive were it to be presented to the CNS as a potential force control model. It would likely require a vast circuitry to control the brief disfacilitation of previouslyrecruited motor units. The organization of the neuromuscular system integrates motor unit recruitment and rate coding mechanisms to resolve the muscle force gradation control problem segmentally, therefore the control demand of the CNS is eased. Based on a fixed recruitment scheme, the anticipated new twitch tension is fused with the active units’
652
G. Kamen and D. C. C. Du
summated forces by asynchronous activation. Further development of appropriate firing rate estimates such as the boxcar method will aid our understanding of motor unit discharge behavior.
Acknowledgements—This study was supported in part by the University of Massachusetts Graduate Research Fund. We are grateful to the subjects who volunteered to participate in this investigation.
REFERENCES
1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34.
Berger R. D., Akselrod S., Gordon D. and Cohen R. J. (1986) An efficient algorithm for spectral analysis of heart rate variability. IEEE Trans. biomed. Eng. BME 33, 900–904. Binder-Macleod S. A. and Clamann H. P. (1989) Force output of cat motor units stimulated with trains of linearly varying frequency. J. Neurophysiol. 61, 208–217. Botterman B. R., Iwamoto G. A. and Gonyea W. J. (1986) Gradation of isometric tension by different activation rates in motor units of cat flexor carpi radialis muscle. J. Neurophysiol. 56, 494–506. Broman H., De Luca C. J. and Mambrito B. (1985) Motor unit recruitment and firing rates interaction in the control of human muscles. Brain Res. 337, 311–319. Bu¨dingen H. J. and Freund H.-J. (1976) The relationship between the rate of rise of isometric tension and motor unit recruitment in a human forearm muscle. Pflugers. Arch. 362, 61–67. Clamann H. P. and Schelhorn T. B. (1988) Nonlinear force addition of newly recruited motor units in the cat hindlimb. Muscle Nerve 11, 1079–1089. de Boer R. W., Karemaker J. M. and Strackee J. (1985) Spectrum of a series of point events, generated by the integral pulse frequency modulation model. Med. Biol. Eng. Comput. 23, 138–142. Feinstein B., Lindegard B., Nyman E. and Wohlfart G. (1955) Morphologic studies of motor units in normal human muscles. Acta anat. 23, 127–142. Fleshman J. W., Munson J. B., Sypert G. W. and Friedman W. A. (1981) Rheobase, input resistance and motor-unit type in medial gastrocnemius motoneurons in the cat. J. Neurophysiol. 46, 1326–1338. Freund H. J. (1983) Motor unit and muscle activity in voluntary motor control. Physiol. Rev. 63, 387–436. Freund H. J., Budingen H. J. and Dietz V. (1975) Activity of single motor units from human forearm muscles during voluntary isometric contractions. J. Neurophysiol. 38, 933–946. Harrison P. J. (1983) The relationship between the distribution of motor unit mechanical properties and the forces due to recruitment and to rate coding for the generation of muscle force. Brain Res. 264, 311–315. Henneman E. (1981) Recruitment of motoneurons: the size principle. In Prog. Clin. Neurophysiol. (ed. Desmedt J. E.). Karger, Basel, New York. Henneman E. and Olson C. B. (1965) Relations between structure and function in the design of skeletal muscle. J. Neurophysiol. 28, 581–598. Henneman E., Somjen G. and Carpenter D. O. (1965) Excitability and inhibitibility of motoneurons of different sizes. J. Neurophysiol. 28, 599–620. Henneman E., Somjen G. and Carpenter D. O. (1965) Functional significance of cell size in spinal motoneurons. J. Neurophysiol. 28, 560–580. Hultborn H. and Pierrot-Deseilligny E. (1979) Changes in recurrent inhibition during voluntary soleus contractions in man studied by an H-reflex technique. J. Physiol. 297, 229–251. Kamen G. and De Luca C. J. (1989) Motor unit firing behavior in the aged. In Computer Aided Electromyography and Expert Systems (ed. Desmedt J. E.), pp. 231–239. Elsevier Science Publishers, New York. Kamen G., Sison S. V., Du D. C. C. and Patten C. (1995) Motor unit discharge behavior in older adults during maximal effort contractions. J. appl. Physiol. 79, 1908–1913. Kernell D. (1992) Organized variability in the neuromuscular system: a survey of task-related adaptations. Archs ital. Biol. 130, 19–66. Kernell D., Eerbeek O. and Verhey B. A. (1983) Relation between isometric force and stimulus rate in cat’s hindlimb motor units of different twitch contraction time. Expl Brain Res. 50, 220–227. Kernell D. and Sjoholm H. (1975) Recruitment and firing rate modulation of motor unit tension in a small muscle of the cat’s foot. Brain Res. 98, 57–72. Kudina L. P. and Alexeeva N. L. (1992) After-potentials and control of repetitive firing in human motoneurones. Electroenceph. clin. Neurophysiol. 85, 345–353. Kudina L. P. and Pantseva R. E. (1988) Recurrent inhibition of firing motoneurones in man. Electroenceph. clin. Neurophysiol. 69, 179–185. Kukulka C. G. and Clamann H. P. (1981) Comparison of the recruitment and discharge properties of motor units in human brachial biceps and adductor pollicis during isometric contractions. Brain Res. 219, 45–55. Lind A. R. and Petrofsky J. S. (1978) Isometric tension from rotary stimulation of fast and slow cat muscles. Muscle Nerve 1, 213–218. Macefield V. G., Gandevia S. C., Bigland-Ritchie B., Gorman R. B. and Burke D. (1993) The firing rates of human motoneurones voluntarily activated in the absence of muscle afferent feedback. J. Physiol. 471, 429–443. Masakado Y., Akaboshi K., Nagata M., Kimura A. and Chino N. (1995) Motor unit firing behavior in slow and fast contractions of the first dorsal interosseous muscle of healthy men. Electroenceph. clin. Neurophysiol. 97, 290–295. Milner-Brown H. S. and Stein R. B. (1973) The orderly recruitment of human motor units during voluntary isometric contractions. J. Physiol. 230, 359–370. Milner-Brown H. S., Stein R. B. and Yemm R. (1973) Changes in firing rate of human motor units during linearly changing voluntary contractions. J. Physiol. 230, 371–390. Monster A. W. (1979) Firing rate behavior of human motor units during isometric voluntary contraction: relation to unit size. Brain Res. 171, 349–354. Paulin M. G. (1992) Digital filters for firing rate estimation. Biol. Cybern. 66, 525–531. Person R. S. and Kudina L. P. (1972) Discharge frequency and discharge pattern in human motor units during voluntary contractions of muscle. Electroenceph. clin. Neurophysiol. 32, 471–483. Petajan J. H. (1991) AAEM minimonograph #3: motor unit recruitment. Muscle Nerve 14, 489–502.
Rate modulation and precision force control 35. 36. 37. 38. 39. 40. 41. 42.
653
Petajan J. H. and Phillip B. A. (1969) Frequency control of motor unit action potentials. Electroenceph. clin. Neurophysiol. 27, 66–72. Petrofsky J. S. (1979) Sequential motor unit stimulation through peripheral motor nerves in the cat. Med. Biol. Eng. Comput. 17, 87–93. Powers R. K. and Binder M. D. (1985) Distribution of oligosynaptic group I input to the cat medial gastrocnemius motoneuron pool. J. Neurophysiol. 53, 497–517. Romaiguere P., Vedel J.-P. and Pagni S. (1989) Fluctuations in motor unit recruitment threshold during slow isometric contractions of wrist extensor muscles in man. Neurosci. Lett. 103, 50–55. Romaiguere P., Vedel J.-P. and Pagni S. (1993) Comparison of fluctuations of motor unit recruitment and de-recruitment thresholds in man. Expl Brain Res. 95, 517–522. Scharf B. (1965) Experimental Sensory Psychology. Scott, Foresman and Co., Glenview, IL. Tokizane T. and Shimazu H. (1964) Functional Differentiation of Human Skeletal Muscle. University of Tokyo Press, Tokyo. Windhorst U. (1988) How Brain-like is the Spinal Cord? Interacting Cell Assemblies in the Nervous System. Springer–Verlag, Berlin. (Accepted 20 April 1998)
APPENDIX
Grounded on the Integral Pulse Frequency Modulation7 model, a modified triangular H2 filtering algorithm of rate estimate has been proposed by Berger and his colleagues.1 This algorithm has been used in heart rate variability studies to derive the heart rate signal from the electrocardiogram. The technique developed by Berger et al.1 estimates the firing rate from a series of impulses by taking the number of ISIs that fall within the local window centered at the sampling time and then dividing by the width of the sampling window. The sample values of these constant time windows are equivalent to the convolution of a stepwise continuous firing rate signal in which the reciprocal of the ISI is carried between two consecutive impulses with a rectangular window that is two sampling intervals wide. Figure 11 presents a schematic description of the algorithm. We applied this algorithm to estimate instantaneous motor unit firing rate using a 100-ms window, equivalent to re-sampling the firing time series at 10 samples/s with a constant time between each sample. We then calculated how many ISIs were enclosed within this time interval. The resultant firing rate estimate includes all the complete ISIs as well as the fractional ISI ratios from both ends of the window boundaries. In cases where the window width only takes a portion of one whole ISI, the ratio of the window width versus the ISI duration is used (as shown in Fig. 11, FR3 has 0.87 of the whole ISI in the sampling window). The calculation for the boundary fractional ISI ratios is the same, i.e. the ratio of the ISI duration within the window versus the whole ISI duration is used. The calculated fractional number of ISI counts was then converted to a firing rate (in pulses/s.)
Fig. 11. Schematic diagram for the algorithm of a modified boxcar filter (from Berger et al.1). There are four windows illustrated, they are labeled as FR1 through FR4. The window width is T for all four samples. The firing rate estimate is calculated as the total ISI count divided by the time (T=100 ms).
PII:
Neuroscience Vol. 88, No. 2, pp. 643–653, 1999 Copyright 1998 IBRO. Published by Elsevier Science Ltd Printed in Great Britain. All rights reserved 0306–4522/99 $19.00+0.00 S0306-4522(98)00248-6
INDEPENDENCE OF MOTOR UNIT RECRUITMENT AND RATE MODULATION DURING PRECISION FORCE CONTROL G. KAMEN* and D. C. C. DU Department of Exercise Science, Totman Building, University of Massachusetts, Amherst, MA 01003, U.S.A. Abstract––The vertebrate motor system chiefly employs motor unit recruitment and rate coding to modulate muscle force output. In this paper, we studied how the recruitment of new motor units altered the firing rate of already-active motor units during precision force production in the first dorsal interosseous muscle. Six healthy adults performed linearly increasing isometric voluntary contractions while motor unit activity and force output were recorded. After motor unit discharges were identified, motor unit firing rates were calculated before and after the instances of new motor unit recruitment. Three procedures were applied to compute motor unit firing rate, including the mean of a fixed number of inter-spike intervals and the constant width weighted Hanning window filter method, as well as a modified boxcar technique. In contrast to previous reports, the analysis of the firing rates of over 200 motor units revealed that reduction of the active firing rates was not a common mechanism used to accommodate the twitch force produced by the recruitment of a new motor unit. Similarly, during de-recruitment there was no tendency for motor unit firing rates to increase immediately following the cessation of activity in other motor units. Considerable consistency in recruitment behavior was observed during repeated contractions. However, firing rates during repeated contractions demonstrated considerably more fluctuation. It is concluded that the neuromuscular system does not use short-term preferential motor unit disfacilitation to effect precise regulation of muscular force output. 1998 IBRO. Published by Elsevier Science Ltd. Key words: motoneuron, motor unit, recruitment, rate coding, firing rate.
Two principal mechanisms have been identified to account for the ability of the muscle to vary its force output. One mechanism involves the recruitment of new motor units according to increasing size—the ‘‘size principle’’.13–16 By increasing the number of active motor units within the motoneuron pool, the total force output increases. Aside from the recruitment of new motor units, the major mechanism by which muscular force is graded involves the modulation of motor unit discharge rates—rate coding.30 Considerable research has supported the size principle of motor unit recruitment, however, our knowledge of the factors that control discharge frequency is not nearly as developed. It is known that an approximately sigmoid relationship exists between firing frequency and force output.3,21 For long-term effects, motor unit discharge rate increases with increasing force output,2,30,31 although during relatively long contractions it decreases when constant force is maintained.33 However, the firing rate of human tibialis anterior muscle shows no progressive decline when *To whom correspondence should be addressed. Abbreviations: FDI, first dorsal interosseous; ISI, interspike interval; MVC, maximum voluntary isometric contraction.
the maximal voluntary contraction is sustained during acute deafferentation.27 Units from similar motoneuron pools tend to exhibit similar fluctuations in their firing rates.18 In addition, there are many short-term factors that affect the relationship between firing frequency and force output, such as membrane afterhyperpolarization23 and recurrent inhibition.17,24 However, the role that rate coding might play in compensating for these short-term influences on muscular force is not fully understood. Many studies have been conducted to investigate either recruitment or firing rate modulation in the motor control system, however, few studies have considered how firing rate modulation and motor unit recruitment interact to produce smooth muscular force. In human voluntary contractions, Broman et al.4 described an interesting phenomenon that suggested an alternative mechanism by which subtle changes in muscular force could be produced by the human motor system. Their experiment was conducted on the tibialis anterior and the first dorsal interosseous muscles. Subjects were asked to produce isometric contractions involving gradual increases in muscular force. When new motor units were recruited, the firing rates of the previously-active motor units were slowed. They suggested that the neuromuscular system was capable of disfacilitating
643
644
G. Kamen and D. C. C. Du
active motor units in order to lessen the incremental force contribution from an entire new motor unit and increase the smoothness of the total force output. However, the idea that individual motoneurons can be selectively disfacilitated is a bit disturbing, as it requires a considerable increase in control complexity. Indeed, a key feature of the size principle is its ability to present a mechanism by which the CNS is freed from the task of selecting a different population of motor units depending on force requirements. The present study was designed to further investigate the interaction between motor unit recruitment and rate coding during slowly-increasing isometric contractions requiring precision force production. If the goal is to produce a finely graded force output, is firing rate reduction a viable means of blending the contribution of additional twitch force from a newlyrecruited motor unit with the summated forces produced by the active units? We studied the influence of recruitment on firing rate modulation by tracking many active motor unit firing rates simultaneously during new unit recruitment. Using intramuscular recording of motor unit activities from the first dorsal interosseous muscle (FDI), we computed the firing rates immediately before and immediately after the recruitment of new motor units when the force output was smoothly increasing. These pre- and postrecruitment firing rates were then statistically compared to study possible inhibitory effects from the newly recruited motor unit to the firing rate of previously active motor units. We also investigated the interaction between motor unit de-recruitment and rate coding during slowly-decreasing isometric contractions during precision force production. The firing rates immediately before and immediately after the de-recruitment of a high threshold unit were calculated. The possible compensatory effects were then examined by comparing statistically the pre- and post- de-recruitment firing rates. EXPERIMENTAL PROCEDURES
Subjects Six adult subjects (20–24 years) consented to participate in the study. These individuals had no known history of neuromuscular disorders. All experimental procedures described were approved by the University Institutional Review Board. Procedures Subjects were seated comfortably at a table with their forearm and hand secured in a specialized apparatus designed to measure isometric force during index finger abduction. Maximum voluntary isometric contractions (MVCs) were measured while the subject was instructed to abduct the index finger at maximal effort. A visual trajectory was displayed on a computer monitor for two tasks. One task required the subject to slowly increase the isometric force at a rate of 4% MVC/s up to 30% MVC, then reduce the force at the same rate to resting level. The other task was similar, except that the range of force was between 10% and 40% MVC. The visual trajectory displayed on the
monitor preceeded the subject’s force by 0.25 s and progressed at the same speed as the subject’s trajectory. This real-time display facilitated subject’s force control to yield precise changes in muscular force. Each triangular trajectory was traced by the subject for four consecutive times during the same recording. A single trial then, consisted of four triangular trajectory contractions. Familiarization All subjects were given at least three days of practice to become familiar with the trajectories. Prior to motor unit recording on the fourth day, an evaluation was conducted to verify the proficiency of trajectory production. The experiment was conducted only when the subject could follow the desired trajectory and keep the computed mean square error within a 3% MVC range. Figure 1 shows a typical force recording and the prototype trajectory is superimposed. Force recordings During isometric contractions, the force signal was acquired using a strain gauge transducer (Interface, model MB-10). The force signal was d.c. amplified, filtered (d.c.=1 kHz), and stored on an instrumentation recorder (TEAC MR-30) for later digitization at 50 samples/s. Motor unit recordings The technique for recording and identifying motor unit activity has been described recently.19 Following the assessment of maximal force production, a four-wire needle electrode was inserted into the belly of the first dorsal interosseous, the index finger abductor. This needle electrode consists of a small diameter (25-gauge) stainless steel cannula containing four 50 µm platinum-iridium wires emerging from a side port 7.5 mm from the tip of the electrode. The wires in the side port were arranged in a square array with approximately 200 µm on each side. The signals obtained from the needle electrode were fed through three separate differential amplifiers (Dantec Counterpoint, input impedance: 200 MÙ, bias current: 25 pA), bandpass filtered (1–10 kHz; 3 dB), and recorded on the same instrumentation recorder (TEAC MR-30) for later digitization at 51,200 samples/s. We also manipulated the location of the needle electrode to sample different motor units in the muscle. For each subject, motor units were sampled from four to six different sites. There were usually three to six distinct motor units that could be detected in each contraction. After signal digitization, an epoch detection program algorithm separated the motor unit action potentials from the baseline signals. These three-channel motor unit signals were then processed by a semi-automatic microcomputerbased procedure which identified individual motor unit discharges based on their activity shapes in all three channels. After motor unit spikes were recognized, the firing information of each motor unit action potential train was extracted in order to obtain firing rate estimates. A sample of the motor unit firing activities obtained with the associated force output is illustrated in Fig. 2. Firing rate estimates In order to evaluate the influence of motor unit recruitment on the firing rate of active motor units, a firing rate estimate method is needed. Many reports have used unique, customized estimate procedures for motor unit firing rate extraction to address specific research questions.22,32,33 However, there is no consensus regarding the optimal procedure to estimate motor unit firing rate. Differences in the procedures used to calculate firing rate can lead to marked discrepancies in the estimated values of a unit’s firing rate. One recent firing rate estimate procedure used
Rate modulation and precision force control
645
Fig. 1. The prototype force trajectory of the experiment is superimposed with one typical force production during the study. The rate of force change was 4% MVC/s for both force increasing and decreasing ramps.
Fig. 2. Motor unit discharge history and the recorded force trace (solid, heavy line) in one segment of slow, increasing isometric contraction.
the reciprocal of the mean of five consecutive interspike intervals (ISIs).19 However, the summation of five ISIs in the human FDI at the 30–40% MVC values in this experiment would yield a window size of approximately 300 ms, and possibly as long as 800 ms. This time window might be too long to detect any short-term firing rate changes due to new motor unit recruitment. A fixed length weighted Hanning window such as the 800 ms duration window used by Broman et al.4 might also be too long to detect any short-term influence of new motor unit recruitment on
motor unit firing rates. Additionally, an 800 ms Hanning window can prolong the duration and consequently mask the appearance of short-term firing rate variations in the time domain. A similar problem is obtained with the use of moving average estimates of firing rate due to the fact that the moving average is a rectangular filter.41 We chose to use a modified boxcar algorithm, developed for cardiac heart rate variability analysis (from Berger et al.,1 see Appendix), for the short-term firing rate estimate. The selected boxcar technique allows for the accurate
646
G. Kamen and D. C. C. Du
Fig. 3. Motor unit recruitment threshold consistency in repeated contractions. The motor unit recruitment threshold in one contraction was plotted against that of the next contraction. The same motor unit was recruited at a similar force level regardless of its recruitment threshold.
estimate of firing rate during short durations comparable to one ISI. In order to make comparison with previous reports, the firing rates immediately before and immediately after the new motor unit recruitment were also calculated by applying both ISI averaging (the reciprocal of the mean of five ISIs) and Hanning window firing rate estimates, using a filter width of 800 ms. RESULTS
Consistency of motor unit recruitment thresholds The consistency of motor unit recruitment thresholds was assessed for all motor units available for analysis. The recruitment threshold was taken as the force score corresponding to the time of the first action potential. From repeated contractions, the recruitment thresholds of motor units observed during one contraction were compared with those of the next contraction whenever these same motor units were available. The scattergraph in Fig. 3 shows that the motor unit recruitment thresholds of one contraction were comparable to these same motor unit recruitment thresholds of the next contraction trial during the same trial. There was a linear correlation (r2 =0.90, n=226) between the recruitment threshold of one contraction and that of the same motor unit in the next contraction. We also inspected the motor unit recruitment thresholds during repeated contractions when the force production was varied in ranges of either 0% to 30% or 10% to 40% MVC. The statistical analysis of Student’s t-test revealed that there were no significant differences of the recruitment thresholds between contractions conducted in either the 0% to 30% or the 10% to 40% MVC force range (n=85; t=1.35; P=0.18). Motor unit discharge rates during new unit recruitment The data selected included 47 contractions in which 256 pairs of motor units (an active motor unit and a newly-recruited motor unit) were identified for
Fig. 4. Motor unit firing rates estimated by the ISI averaging method. (A) The scatter distribution of pre- and postrecruitment motor unit firing rates. (B) The relationship between firing rate difference at the new motor unit recruitment and the threshold difference between the active and newly-recruited motor units.
analysis. In these 256 cases, five consecutive ISIs of the active motor unit were available for analysis both before and after new motor unit recruitment. We also recorded 169 cases in which the motor unit was firing for at least 800 ms when a new motor unit was recruited, and maintained its firing for at least 800 ms after recruitment. These 169 sets of ISI data from 47 contractions were used for the weighted Hanning window estimate of firing rate (filter width=800 ms). Motor unit firing rates before and after new motor unit recruitment were computed using the ISI averaging (Fig. 4A) and the Hanning window filter method (Fig. 5A). As illustrated in these scattergrams, there was no consistent change in the firing rate of active motor units when a new motor unit was recruited. While some motor units did decrease firing rate when other motor units were recruited, there were also many units that exhibited an increase in firing rate. The differences between post- and prerecruitment firing rates are plotted as a function of the differences in recruitment threshold in Figs 4B and 5B using the ISI averaging estimation and Hanning window filter algorithm, respectively. These figures indicate that similar trends in the data were obtained irrespective of the method used to compute firing rate.
Rate modulation and precision force control
647
Fig. 5. Motor unit firing rates evaluated by the weighted Hanning window filter. (A) The relation between pre- and post-recruitment motor unit firing rates. (B) The motor unit firing rate changes against the corresponding threshold differences.
Fig. 6. Motor unit firing rate estimates derived by a modified boxcar algorithm. (A) The distribution of motor unit firing rates evaluated before and after recruitment. (B) The relationship between the firing rate changes and the recruitment threshold differences.
Statistical analysis of firing rate changes
Short-term discharge rate estimate using the boxcar algorithm
Using the reciprocal of the mean of five consecutive ISIs, the mean firing rate of the already-active motor units was 0.254.403 pulses/s (n=256) faster following the recruitment of a new motor unit, and a paired Student’s t-test revealed no significant difference in firing rate (t=0.89; P=0.37). From the Hanning window filtering calculations, the firing rate difference immediately following new recruitment was 1.965.262 pulses/s (n=169), and this increase in firing rate was statistically significant (t=4.85; P<0.05). Thus, rather than obtaining a decrease in firing rate, both of these analyses revealed an increase in firing rate, likely attributable to the requirements of the increased force demand over the relatively long time-course used in these two analyses. The observation that the Hanning window calculation was almost eight times greater than the reciprocal of the five ISIs (1.96 pps vs 0.25 pps) supports our concern that the 800 ms Hanning window uses a time-course that is too long to be sufficiently sensitive to shortterm disfacilitation of previously-active motor units. Firing rate estimates obtained with this Hanning window appear to be considerably more sensitive to longer-term increases in force production than short-term disfacilitation effects.
From the same 47 contractions, we were able to obtain 266 instances that a motor unit was firing at least 100 ms before a new motor unit was recruited, and the activity continued for at least 100 ms. We applied the modified boxcar procedure as described in the Appendix to these 266 motor units to compute the instantaneous firing rates before and after new motor unit recruitment using an interval of 100 ms. The motor units in the FDI muscle begin firing at roughly 10 pulses/s at recruitment.11,30,33,41 Motor unit firing rates in this muscle are normally in the range of eight to 40 pulses/s when the rate of force production is moderate and the maximal force level of the contraction does not exceed 50% MVC.19 Consequently, a short-term firing rate evaluation interval of 100 ms should be sufficient to encompass short duration ISI variations when firing rates fall into these ranges. If new motor unit recruitment causes only a few ISIs to be lengthened after the moment of recruitment, the firing rate estimate using the shorter 100-ms window will be a more powerful tool for differentiating this short-term variation of motor unit firing rate. The results of the boxcar analysis are shown in Fig. 6. The pre- and post-recruitment motor unit firing
648
G. Kamen and D. C. C. Du
Fig. 7. Active motor unit pre-recruitment firing rate in repeated contractions. Each motor unit’s firing rate estimated before a new motor unit was recruited was plotted against the same unit’s firing rate estimate before the same new motor unit was recruited again in the next contraction.
Fig. 8. Motor unit firing rate variations. In repeated contractions from the same subject, the change in firing rate for five different motor units are presented over the course of four consecutive contractions in two different force ranges.
rates were approximately evenly distributed along the line describing x=y. When the motor unit firing rates before and after new motor unit recruitment were compared by paired Student’s t-test, the results showed that there were no significant differences between the post- and pre-recruitment firing rates with the mean of the differences equal to 0.373.842 pulses/s (n=266; t=1.55; P=0.12).
ous ramp force productions in which force varied from 0% to 30% MVC. The sequences of H1 to H4 were drawn from the four consecutive force productions executed within the 10%–40% MVC range. There were five distinct motor units identified in each of the four ramp force contractions, and these same five units were also identified in the two different force ranges. The motor units were numbered according to the order of recruitment thresholds, i.e. the smaller motor unit numbers were given to the lower threshold units. When the same new motor unit was recruited, the firing rate changes varied considerably for all five motor units. At the first ramp force production of the 0%–30% MVC range, all five motor units increased their firing rates at new unit recruitment. However, in the last ramp force production of the 10%–40% MVC range, these five motor units decreased their firing rates when the same new unit was recruited again. In other ramp contractions, as can be seen from the figure, some motor units increased their firing rates while other motor unit firing rates decreased at the time of recruitment. In addition, the direction of firing rate change in repeated contractions varied considerably, regardless of the recruitment threshold and the range of force production. The sample illustrated in Fig. 8 was selected on the basis of the relatively large number of motor units (five) available for analysis from the same subject during these consecutive contractions.
Firing rate analysis of repeated contractions The repeatability of motor unit firing rate estimates obtained immediately prior to the recruitment of new motor units (‘‘recruitment rate’’34,35) was also assessed using the repeated contractions. The modified boxcar algorithm was used for the remainder of the firing rate estimates in the following analyses. During the repeated contractions, an active motor unit whose firing rate was estimated immediately prior to the recruitment of a new motor unit was examined in the next contraction at the precise point in the isometric force recording where the newlyactive unit was again recruited. These two motor unit firing rate estimates of the same unit in consecutive contractions are plotted in Fig. 7. Repeated firing rate estimates revealed a moderate correlation (r2 =0.54, n=226) with considerable dispersion. We also scrutinized the stability of active motor unit firing rate changes at the moment of new unit recruitment in repeated contractions. There was no observable consistency of motor unit firing rate change in repeated contractions. A sample of these firing rate changes in the consecutive contractions is shown in Fig. 8 where the firing rate changes are the differences between post- and pre-recruitment firing rates. Two different force ranges are illustrated, including one contraction conducted within 0% to 30% MVC and the other executed between 10% and 40% MVC. In each contraction, there were four sequential ramp force variations. In the figure, the series labeled L1 through L4 were the four continu-
De-recruitment analysis The relationship between motor unit recruitment and de-recruitment behavior was investigated in the same ramp contraction. The force values corresponding to the instances of the first and the last action potential were used as the recruitment and de-recruitment thresholds, respectively. The statistical analysis of the recruitment threshold and the de-recruitment threshold revealed a high Pearson correlation (r2 =0.81; n=263). The inter-relationship
Rate modulation and precision force control
649
Fig. 9. The scattergram of recruitment threshold versus de-recruitment threshold. There were significant correlations between recruitment and de-recruitment thresholds in one ramp contraction.
between recruitment and de-recruitment thresholds is illustrated in Fig. 9. To what extent might the firing rate of active units be altered when a high-threshold motor unit was de-recruited? We observed more than 200 pairs of motor units when the active motor units were still firing and maintained their activities for at least the interval required by the firing rate estimate method (boxcar algorithm, width=100 ms) at the highthreshold motor unit de-recruitment. Firing rate estimates were computed before and after the instances of motor unit de-recruitment. The results (see Fig. 10A) indicated that motor unit firing rates had a tendency to decrease continuously when a high-threshold unit was de-recruited: mean postde-recruitment vs pre-de-recruitment firing rate was 1.153.647 pulses/s (n=285). Differences between post- and pre-de-recruitment firing rate were compared by paired Student’s t-test which was statistically significant (t=5.32; P<0.05). One might suspect that the tendency to change firing rate prior to de-recruitment would be greatest among motor units with the greatest difference in recruitment thresholds. However, the results showed that there was no correlation between the change in firing rate and the relative change in recruitment threshold (see Fig. 10B). DISCUSSION
The results of this investigation refute the idea that the neuromuscular control system compensates for new motor unit recruitment by disfacilitating the active units in order to maintain precision during force production. Motor unit firing rates of alreadyactive motor units were estimated using several procedures, including the implementation of a boxcar algorithm (see Appendix) to ensure accuracy of short-term firing rate estimates. There are several reasons why the data presented obviate the need to postulate the existence of motoneuron disfacilitation while neighboring motoneurons are recruited.
Fig. 10. The firing rate variation at motor unit derecruitment. (A) The scattergram of motor unit firing rates estimated before and after de-recruitment. (B) The relationship between firing rate change and the difference of recruitment thresholds.
The problem of force regulation The CNS is assigned the task of smoothly grading force output by jointly controlling the numbers of active motor units and their respective firing rates. However, precision force tasks require the integration of individual twitch forces generated by all active motor units to produce a continuous, smooth contraction. So the problem faced by the CNS requires the translation of randomly-active digital units into a smooth analogue output. The force gradation problem is made more arduous when sufficient extra force is required to justify recruitment of an additional motor unit. At the moment of new motor unit recruitment, the additional twitch force has to be blended with the force production of the active units. Both the intensity of this new twitch force and the timing of the twitch with respect to the already-firing motor units have to be accommodated to result in a smooth, finelygraded force output. In this study, we investigated this aspect of the neuromuscular control mechanism, the influence of motor unit recruitment on active motor unit firing rates, when the muscle force output was slowly increased during a linear ramp trajectory. The influences of the rate of tension development The rate of isometric tension development influences motor unit recruitment as well as firing rate
650
G. Kamen and D. C. C. Du
modulation. The motor unit firing rate increases as the speed of the contraction increases.28 All motor units have lower recruitment thresholds with increases in the speed of force production.5,28 Although the increasing rate of isometric tension development lowers the recruitment threshold of motor units, the mechanical twitch tension from the newly recruited motor unit contributes at the same force level regardless of the speed of contraction.5 Whenever a new motor unit is recruited, the mechanism of force regulation adapts to the additional twitch tension in the same way throughout the whole range of force production. With this invariant recruitment framework, the control mechanism of the neuromuscular system can be simplified without addressing each additional motor unit twitch force differently. The force level at the onset of recruitment was used as the recruitment threshold. To minimize the effect of lowering motor unit recruitment threshold caused by the speed of contraction, we explored the minimal rate of force production that subjects could perform and 4% MVC/s was the slowest rate of force change not associated with moderate force variation. It is known that motor unit recruitment thresholds do not change when the contraction speed is sufficiently slow.10 Additionally, the maximal number of distinct motor units can be recruited successively only when a precision, isometric contraction is conducted at an extremely slow rate. However, only small numbers of motor units are recruited and they fire slowly at low force level. Due to the limited number of twitch tensions being fused, an increasing force variation is often observed at these low forces.34 This intrinsic property of motor unit twitch tension, one discrete twitch for each firing, might well define the minimal rate of force development. Are additional force regulatory mechanisms necessary? By contrasting the newly recruited unit’s tension with the summation of the smaller units’ force output, Henneman and Olson14 indicated that the twitch contraction from the newly-activated motor unit added tension at a constant proportion to the total cumulative force. Milner-Brown and Stein29 also observed this approximate constancy of relative recruitment steps in the first dorsal interosseous muscle. They observed that the twitch tension of the new motor unit varied nearly linearly as a function of the level of voluntary force at which the unit was recruited. Thus, it is possible that this constancy between force requirement and current force level obviates the need to theorize the existence of additional force regulatory mechanisms. Is compensatory disfacilitation an operative mechanism? If compensatory disfacilitation were an active control mechanism, there are several reasons it should
have been observed in the present study designed to reveal such a mechanism. First, small muscles rely more on firing rate modulation than larger muscles.25 Any phenomenon that produces momentary fluctuations in firing rate should be especially apparent in the small FDI muscle chosen for this investigation. Second, firing rate modulation becomes a more important mechanism in force graduation at intermediate force levels.30 The force levels varying up to 40% MVC should have optimized the use of compensatory disfacilitation during new recruitment. Third, compensatory disfacilitation should be especially attractive in a muscle with smaller numbers of motor units; the first dorsal interosseous muscle contains approximately 120 motor units.8 Small numbers of motor units would exacerbate the digital-to-analogue translation problem. Fourth, the task of increasing and decreasing force at a rate of 4% MVC/s requires considerable precision. This gradual increase in force and rate of motor unit recruitment should emphasize the rate coding mechanism and serve as an ideal environment for a mechanism such as compensatory disfacilitation. Finally, the boxcar method of computing discharge rate should offer the greatest sensitivity to identify any momentary rate changes. Temporary cessation of motor unit activity has been observed in some situations. For example, when a motor unit is recruited during a small but rapid increase in muscular force, it often stops firing when the force level reaches a new stable value and resumes continuous discharge only when the force output reaches a new tonic level.11 We do know that the increased rate of force production results in earlier motor unit recruitment with increasing firing rate.5,28 After a short force increase, it is possible that reduction of the supraspinal presynaptic excitation to the motoneuron pool produces a cessation of highthreshold unit firing and a reduction of active unit firing during the subsequent plateau contraction. Perhaps this impulsive disfacilitation was observed by Broman et al.4 when they asked subjects to hold the level of force constant immediately after recruitment. The lengthened motor unit ISIs could have resulted from the abrupt change in the rate of force production. Repeated contractions The results demonstrated that recruitment thresholds were stable in repeated contractions in which the rate of tension increase was kept constant. This is consistent with the results from other reports.10,34,38 Moreover, motor units were also recruited at similar levels of force production irrespective of the range of contraction. On the other hand, the active motor unit firing rates at the moment before a new motor unit was recruited were distributed in a wider range than the motor unit recruitment thresholds. It has been suggested that recruitment provides a coarse force gradation, while firing rate modulation complements
Rate modulation and precision force control
the control mechanism by providing a finer gradation of muscle contraction.34 Based on the invariant recruitment framework, the smallest motor unit is the first unit activated, the magnitude of the tension contribution and the moment when the twitch tension will be fused with the total force production can be predicted from the size of the unit. After motor unit recruitment, the rate coding mechanism controls the tension output of the unit. With a wide range of firing rate, the force contribution of the unit can be finely modulated.
De-recruitment In congruence with previous observations, a positive relationship (r2 =0.81) was observed between motor unit recruitment threshold and de-recruitment threshold.39 Motor unit firing rate estimates computed before and after de-recruitment yielded similar findings to the recruitment analysis: the cessation of motor unit activity does not alter the firing rate decrement of other motor units. The neuromuscular system controls the cessation of motor unit activity without compensatory control of active motor unit firing rates. The distribution of motor units in the muscle has been found to be highly skewed, having many smaller units and progressively smaller numbers of large units.14,29,37 Newly-activated motor units are approximately constant in size compared to the total tension of already-activated units. It has been suggested that this observation is in accordance with the Weber-Fechner relationship of sensory psychophysics,12,20,40 under which the relationship between the cumulative force (logarithmic) and the number of recruited motor units (linear) is logarithmic. The extension of the Weber-Fechner principle to motor unit recruitment is logical, as there are many smaller motor units recruited at low force levels and fewer units are recruited as the force level increases. Such a logarithmic relationship has indeed been illustrated in cat muscles.3,9 One way to achieve the finest gradation of force production is to maximize the number of motor units. However, this is not the organization used in the neuromuscular control system. Rather, motor units are recruited at different degrees of CNS excitation and the recruitment mechanism activates the smallest units first, ratcheting up to the largest units according to the size principle. This motor unit recruitment scheme, based on the Weber-Fechner principle of sensory psychophysics, was considered by Kernell20 as a possible means through which the neuromuscular system can achieve steady and finely gradable force production using recruitment and rate coding. The number of motor units is minimized so that smooth force gradation is achieved when a new twitch tension is added to the total force production according to the Weber-Fechner principle. Therefore,
651
the motor unit recruitment mechanism itself is furnished with the capability to generate smooth, graded force production. However, recruitment of all the motoneurons of the pool can only produce 10–25% of the maximal tension generated when all the units’ firing rates are at the minimum level to cause twitch tension fusion.10 At higher force levels, motor unit firing rate modulation contributes more to force production than recruitment.30 The neuromuscular system integrates the motor unit twitch tensions by enforcing predominant asynchronous motor unit firing behavior and asynchronous fusion of unit twitch tensions.26,36 Indeed, it has been suggested that one of the effects of Renshaw recurrent inhibition is to enhance the asynchronous firing between active motor units.42 Moreover, there are observations that synchronous activation of adjacent motor units increases the total force output non-linearly by force argumentation.6 The control mechanism of rate coding is far more complex than that of recruitment as illustrated in the results. Firing rates of the active units when new units are recruited varied somewhat in different contractions whereas the recruitment thresholds were consistent during repeated contractions. One possibility is that each motor unit has an inherent discharge variability, including a discharge variability at the onset of recruitment. Such variability could be due, in part, to the spatiotemporal structure of synaptic input to each motoneuron. Although the proportion of recruitment and rate coding is different for each individual muscle group, the demand of finely graded force production and the neuromuscular control mechanisms are similar. From the above discussions, whenever the motor unit recruitment is wellorganized, the condition of smooth force production can be achieved. Superimposed on motor unit recruitment, the rate coding mechanism regulates the firing time of active motor units. However, the manner in which rate coding is regulated so that each twitch is fused for smooth, finely graded force output will require further exploration. CONCLUSIONS
The results of this experiment provide no evidence for the idea that the motor control system compensates for new motor unit recruitment by disfacilitating the active units. Compensatory disfacilitation would be computationally expensive were it to be presented to the CNS as a potential force control model. It would likely require a vast circuitry to control the brief disfacilitation of previouslyrecruited motor units. The organization of the neuromuscular system integrates motor unit recruitment and rate coding mechanisms to resolve the muscle force gradation control problem segmentally, therefore the control demand of the CNS is eased. Based on a fixed recruitment scheme, the anticipated new twitch tension is fused with the active units’
652
G. Kamen and D. C. C. Du
summated forces by asynchronous activation. Further development of appropriate firing rate estimates such as the boxcar method will aid our understanding of motor unit discharge behavior.
Acknowledgements—This study was supported in part by the University of Massachusetts Graduate Research Fund. We are grateful to the subjects who volunteered to participate in this investigation.
REFERENCES
1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34.
Berger R. D., Akselrod S., Gordon D. and Cohen R. J. (1986) An efficient algorithm for spectral analysis of heart rate variability. IEEE Trans. biomed. Eng. BME 33, 900–904. Binder-Macleod S. A. and Clamann H. P. (1989) Force output of cat motor units stimulated with trains of linearly varying frequency. J. Neurophysiol. 61, 208–217. Botterman B. R., Iwamoto G. A. and Gonyea W. J. (1986) Gradation of isometric tension by different activation rates in motor units of cat flexor carpi radialis muscle. J. Neurophysiol. 56, 494–506. Broman H., De Luca C. J. and Mambrito B. (1985) Motor unit recruitment and firing rates interaction in the control of human muscles. Brain Res. 337, 311–319. Bu¨dingen H. J. and Freund H.-J. (1976) The relationship between the rate of rise of isometric tension and motor unit recruitment in a human forearm muscle. Pflugers. Arch. 362, 61–67. Clamann H. P. and Schelhorn T. B. (1988) Nonlinear force addition of newly recruited motor units in the cat hindlimb. Muscle Nerve 11, 1079–1089. de Boer R. W., Karemaker J. M. and Strackee J. (1985) Spectrum of a series of point events, generated by the integral pulse frequency modulation model. Med. Biol. Eng. Comput. 23, 138–142. Feinstein B., Lindegard B., Nyman E. and Wohlfart G. (1955) Morphologic studies of motor units in normal human muscles. Acta anat. 23, 127–142. Fleshman J. W., Munson J. B., Sypert G. W. and Friedman W. A. (1981) Rheobase, input resistance and motor-unit type in medial gastrocnemius motoneurons in the cat. J. Neurophysiol. 46, 1326–1338. Freund H. J. (1983) Motor unit and muscle activity in voluntary motor control. Physiol. Rev. 63, 387–436. Freund H. J., Budingen H. J. and Dietz V. (1975) Activity of single motor units from human forearm muscles during voluntary isometric contractions. J. Neurophysiol. 38, 933–946. Harrison P. J. (1983) The relationship between the distribution of motor unit mechanical properties and the forces due to recruitment and to rate coding for the generation of muscle force. Brain Res. 264, 311–315. Henneman E. (1981) Recruitment of motoneurons: the size principle. In Prog. Clin. Neurophysiol. (ed. Desmedt J. E.). Karger, Basel, New York. Henneman E. and Olson C. B. (1965) Relations between structure and function in the design of skeletal muscle. J. Neurophysiol. 28, 581–598. Henneman E., Somjen G. and Carpenter D. O. (1965) Excitability and inhibitibility of motoneurons of different sizes. J. Neurophysiol. 28, 599–620. Henneman E., Somjen G. and Carpenter D. O. (1965) Functional significance of cell size in spinal motoneurons. J. Neurophysiol. 28, 560–580. Hultborn H. and Pierrot-Deseilligny E. (1979) Changes in recurrent inhibition during voluntary soleus contractions in man studied by an H-reflex technique. J. Physiol. 297, 229–251. Kamen G. and De Luca C. J. (1989) Motor unit firing behavior in the aged. In Computer Aided Electromyography and Expert Systems (ed. Desmedt J. E.), pp. 231–239. Elsevier Science Publishers, New York. Kamen G., Sison S. V., Du D. C. C. and Patten C. (1995) Motor unit discharge behavior in older adults during maximal effort contractions. J. appl. Physiol. 79, 1908–1913. Kernell D. (1992) Organized variability in the neuromuscular system: a survey of task-related adaptations. Archs ital. Biol. 130, 19–66. Kernell D., Eerbeek O. and Verhey B. A. (1983) Relation between isometric force and stimulus rate in cat’s hindlimb motor units of different twitch contraction time. Expl Brain Res. 50, 220–227. Kernell D. and Sjoholm H. (1975) Recruitment and firing rate modulation of motor unit tension in a small muscle of the cat’s foot. Brain Res. 98, 57–72. Kudina L. P. and Alexeeva N. L. (1992) After-potentials and control of repetitive firing in human motoneurones. Electroenceph. clin. Neurophysiol. 85, 345–353. Kudina L. P. and Pantseva R. E. (1988) Recurrent inhibition of firing motoneurones in man. Electroenceph. clin. Neurophysiol. 69, 179–185. Kukulka C. G. and Clamann H. P. (1981) Comparison of the recruitment and discharge properties of motor units in human brachial biceps and adductor pollicis during isometric contractions. Brain Res. 219, 45–55. Lind A. R. and Petrofsky J. S. (1978) Isometric tension from rotary stimulation of fast and slow cat muscles. Muscle Nerve 1, 213–218. Macefield V. G., Gandevia S. C., Bigland-Ritchie B., Gorman R. B. and Burke D. (1993) The firing rates of human motoneurones voluntarily activated in the absence of muscle afferent feedback. J. Physiol. 471, 429–443. Masakado Y., Akaboshi K., Nagata M., Kimura A. and Chino N. (1995) Motor unit firing behavior in slow and fast contractions of the first dorsal interosseous muscle of healthy men. Electroenceph. clin. Neurophysiol. 97, 290–295. Milner-Brown H. S. and Stein R. B. (1973) The orderly recruitment of human motor units during voluntary isometric contractions. J. Physiol. 230, 359–370. Milner-Brown H. S., Stein R. B. and Yemm R. (1973) Changes in firing rate of human motor units during linearly changing voluntary contractions. J. Physiol. 230, 371–390. Monster A. W. (1979) Firing rate behavior of human motor units during isometric voluntary contraction: relation to unit size. Brain Res. 171, 349–354. Paulin M. G. (1992) Digital filters for firing rate estimation. Biol. Cybern. 66, 525–531. Person R. S. and Kudina L. P. (1972) Discharge frequency and discharge pattern in human motor units during voluntary contractions of muscle. Electroenceph. clin. Neurophysiol. 32, 471–483. Petajan J. H. (1991) AAEM minimonograph #3: motor unit recruitment. Muscle Nerve 14, 489–502.
Rate modulation and precision force control 35. 36. 37. 38. 39. 40. 41. 42.
653
Petajan J. H. and Phillip B. A. (1969) Frequency control of motor unit action potentials. Electroenceph. clin. Neurophysiol. 27, 66–72. Petrofsky J. S. (1979) Sequential motor unit stimulation through peripheral motor nerves in the cat. Med. Biol. Eng. Comput. 17, 87–93. Powers R. K. and Binder M. D. (1985) Distribution of oligosynaptic group I input to the cat medial gastrocnemius motoneuron pool. J. Neurophysiol. 53, 497–517. Romaiguere P., Vedel J.-P. and Pagni S. (1989) Fluctuations in motor unit recruitment threshold during slow isometric contractions of wrist extensor muscles in man. Neurosci. Lett. 103, 50–55. Romaiguere P., Vedel J.-P. and Pagni S. (1993) Comparison of fluctuations of motor unit recruitment and de-recruitment thresholds in man. Expl Brain Res. 95, 517–522. Scharf B. (1965) Experimental Sensory Psychology. Scott, Foresman and Co., Glenview, IL. Tokizane T. and Shimazu H. (1964) Functional Differentiation of Human Skeletal Muscle. University of Tokyo Press, Tokyo. Windhorst U. (1988) How Brain-like is the Spinal Cord? Interacting Cell Assemblies in the Nervous System. Springer–Verlag, Berlin. (Accepted 20 April 1998)
APPENDIX
Grounded on the Integral Pulse Frequency Modulation7 model, a modified triangular H2 filtering algorithm of rate estimate has been proposed by Berger and his colleagues.1 This algorithm has been used in heart rate variability studies to derive the heart rate signal from the electrocardiogram. The technique developed by Berger et al.1 estimates the firing rate from a series of impulses by taking the number of ISIs that fall within the local window centered at the sampling time and then dividing by the width of the sampling window. The sample values of these constant time windows are equivalent to the convolution of a stepwise continuous firing rate signal in which the reciprocal of the ISI is carried between two consecutive impulses with a rectangular window that is two sampling intervals wide. Figure 11 presents a schematic description of the algorithm. We applied this algorithm to estimate instantaneous motor unit firing rate using a 100-ms window, equivalent to re-sampling the firing time series at 10 samples/s with a constant time between each sample. We then calculated how many ISIs were enclosed within this time interval. The resultant firing rate estimate includes all the complete ISIs as well as the fractional ISI ratios from both ends of the window boundaries. In cases where the window width only takes a portion of one whole ISI, the ratio of the window width versus the ISI duration is used (as shown in Fig. 11, FR3 has 0.87 of the whole ISI in the sampling window). The calculation for the boundary fractional ISI ratios is the same, i.e. the ratio of the ISI duration within the window versus the whole ISI duration is used. The calculated fractional number of ISI counts was then converted to a firing rate (in pulses/s.)
Fig. 11. Schematic diagram for the algorithm of a modified boxcar filter (from Berger et al.1). There are four windows illustrated, they are labeled as FR1 through FR4. The window width is T for all four samples. The firing rate estimate is calculated as the total ISI count divided by the time (T=100 ms).