Toe flexor muscle spindle discharge and stretch modulation during locomotor activity in the decerebrate cat


  • P. R. Murphy,

    Corresponding author
    1. Department of Neuroscience, Medical School, University of Newcastle, Newcastle upon Tyne, NE2 4HH, UK
      Corresponding author P. R. Murphy: Department of Neuroscience, Medical School, University of Newcastle, Newcastle upon Tyne NE2 4HH, UK. Email:
    Search for more papers by this author
  • K. G. Pearson,

    1. Department of Physiology and Centre for Neuroscience, University of Alberta, Edmonton, Alberta, Canada T6G 2S2
    Search for more papers by this author
  • R. B. Stein

    1. Department of Physiology and Centre for Neuroscience, University of Alberta, Edmonton, Alberta, Canada T6G 2S2
    Search for more papers by this author

Corresponding author P. R. Murphy: Department of Neuroscience, Medical School, University of Newcastle, Newcastle upon Tyne NE2 4HH, UK. Email:


In order to investigate the nature (i.e. static or dynamic) of fusimotor drive to the flexor hallucis longus (FHL) and flexor digitorum longus (FDL) muscles during locomotion we recorded Ia and group II muscle spindle afferent responses to sinusoidal stretch (0.25 and 1 mm amplitude, respectively, 4–5 Hz) in a decerebrate cat preparation. FHL Ia and group II afferents generally had increased discharge rates and decreased modulation to stretch throughout the step cycle, compared to rest, suggesting raised static γ drive at all locomotor phases. Although the modulation of Ia afferents was reduced during locomotion, most (13 of 18) showed a clear increasing trend during homonymous muscle activity (extension). This was consistent with phasic dynamic γ drive to FHL spindles linked with α drive. In agreement with previous reports, FHL gave a single burst of EMG activity during the step cycle while FDL α drive had two components. One was related to extension while the other comprised a brief burst around the end of this phase. Typically FDL Ia and group II afferents also had elevated firing rates and reduced modulation at all locomotor phases, again implicating static γ drive. Half the afferents (seven Ia, three group II) showed increased discharge during extension, suggesting phasic static γ drive. There was no γ drive associated with the late FDL α burst. In conclusion, the γ drives to FHL and FDL differed during locomotion. FHL, which has the α drive of a classic extensor, received γ drive that closely resembled other extensors. The γ drive of FDL, which exhibits both extensor and flexor α synergies, did not match either muscle type. These observations are compatible with the view that fusimotor drive varies in different muscles during locomotion according to the prevailing sensorimotor requirements.

Group I muscle afferents (Ia and Ib) play an important role in the reflex control of muscle activity during locomotion (for reviews see Pearson, 1995; Zehr & Stein, 1999; Pearson, 2000). Feedback from muscle spindle Ia afferents can be powerfully modulated in the periphery by the action of γ-efferents (fusimotor neurones) which only produce contractions of intrafusal muscle fibres within the spindle and comprise two functionally distinct types, static and dynamic (Matthews, 1981). While both types excite Ia afferents they have opposite effects on responses to muscle stretch. During locomotion γ-efferents are activated but the governing rules are not fully understood (for reviews see Murphy & Martin, 1993; Prochazka, 1996; Murphy, 2000). An important issue concerns the γ drive to different muscles. Perhaps of greatest interest, from a functional standpoint, is the question of how static and dynamic fusimotor actions vary from muscle to muscle. Evidence of muscle-specific γ drive comes mainly from cat preparations during walking. The most striking variation described so far concerns the nature of the strong phasic γ drives of ankle antagonists that are activated with homonymous α activity (α-linked). In tibialis anterior (flexor) static γ discharge is strongly rhythmic (Cabelguen, 1981; Murphy & Hammond, 1993), while in the triceps surae (extensor) it is dynamic γ-efferents that display such behaviour (Perret & Berthoz, 1973; Cabelguen, 1981; Murphy et al. 1984; Taylor et al. 1985; Bessou et al. 1990). Other studies, however, have emphasised the relationship of modulated static γ drive to active muscle shortening during locomotion in both ankle flexor (tibialis anterior) and extensor (medial gastrocnemius) muscles (Taylor et al. 2000a,b).

In the context of muscle-specific γ drive, the locomotor discharge patterns of FHL and FDL γ-efferents were recently described in a decerebrate cat preparation (Murphy, 2002). ‘Tonic’ and ‘phasic’γ patterns were distinguished in both distal hindlimb muscles, whose α activity was similar to that of the intact cat (O'Donovan et al. 1982), but their nature (i.e. static or dynamic) could not be determined as recordings were made from cut nerve filaments. FDL and FHL have a common action on the toes (flexor) but only FHL is a major ankle extensor. Their α-motoneurones possess homonymous, and heteronymous (i.e. between FHL and FDL), monosynaptic Ia connections (Fleshman et al. 1984). Despite these similarities, during locomotion FHL and FDL have very different patterns of activation (O'Donovan et al. 1982; Loeb, 1993; Trank & Smith, 1996). FHL is strongly recruited in a single stereotyped burst with extensors and contributes to anti-gravity support. In contrast, any activity in FDL during extension is generally at a low level. Characteristically this muscle is briefly and weakly recruited with flexors around the end of stance when the toes flex. FHL and FDL provide an interesting test of the specificity of γ drive during walking because of their related anatomical action and ‘Ia synergy’, but distinct usage. In the present study we sought evidence of the nature of the γ locomotor drives to FDL and FHL by recording Ia and group II muscle spindle afferents during sinusoidal stretch (0.25 and 1 mm amplitude, respectively, 4–5 Hz). With these parameters of stretch, static γ activity decreases, while dynamic activity increases Ia afferent stretch modulation (Taylor et al. 1985). We posited that if the nature of γ drive is related to muscle function then differential fusimotor action should occur in FHL and FDL. The aim of the investigation was to test this prediction. A brief account of some of this work has been published in abstract form (Murphy et al. 2002).


Seven adult cats of either sex were used in this series of experiments. All procedures were carried out with approval from University of Alberta Health Sciences Animal Welfare Committee. Anaesthesia was induced in a box using 4 % halothane delivered in 95 % O2 and 5 % CO2. Anaesthesia was then continued using a face mask with 2 % halothane in 95 % O2/5 % CO2 and subsequently, with the same mixture, via a tracheal cannula. The level of anaesthesia was judged from arterial blood pressure recording and by testing the corneal and flexion withdrawal reflexes. Cannulas were inserted in a carotid artery (for monitoring blood pressure) and the jugular vein (for injection of dextran). The right hindlimb was denervated including the hip and tail except for branches to FHL and FDL. A laminectomy was performed to expose L6-S1 dorsal roots. EMGs were recorded using pairs of stainless steel wires (AS632; Cooner Wire, Chatsworth, CA, USA) implanted into the muscle bellies of FDL and FHL. The wires were Teflon-coated except for a 5 mm bared region. The cats were supported with the head in a stereotaxic frame and a clamp on L4 vertebra. The hip was supported with a length of heavy wire that passed through both iliac crests. The right leg was supported by clamps on the knee and foot. The animals were decerebrated, with removal of the forebrain, by transecting the brain stem at a 50 deg angle from the anterior edge of the superior colliculus (premammillary level). This procedure renders the animal totally insentient. Rectal temperature and the temperature of a paraffin pool over the dorsal roots were monitored continuously and maintained near normal body temperature by radiant heat.

A part of the L7 dorsal root was cut and divided to obtain recordings from single muscle spindle afferents to FDL or FHL. Muscle afferents were recorded with an electrode having two platinum wires. The nerves to FDL and FHL were placed in a bipolar cuff (Silastic) electrode that was used to produce twitch contractions and to obtain the axonal conduction latency of afferent fibres. Afferents from Golgi tendon organs and muscle spindles were distinguished by their responses to maximal isometric twitch contractions (Matthews, 1981). Conduction velocity and the responses to ramp stretch and vibration were used to distinguish Ia (FHL: 78–96 m s−1, n/ 18; FDL: 75–96 m s−1, n= 14) and group II (FHL: 37–72 m s−1, n= 10; FDL: 24–62 m s−1, n= 7) spindle afferents.

Superficial muscles were reflected to expose the tendons of FDL and FHL which were cut near the musculo-tendinous junction and tied to two separate servo-controlled muscle pullers. Each puller comprised a strain gauge which was connected to a PMI ServoDisc DC motor (stiffness, 67 N mm−1; PMI Motion Technologies, Commack, NY, USA) incorporating a potentiometer for muscle stretch measurement. Each muscle was slowly stretched and set at a length that produced maximal twitch force during maximal twitch contractions. The superficial muscles were repositioned, covered with skin flaps and the incision closed. Sinusoidal stretches (4-5 Hz, 0.25 mm or 1 mm amplitude) were applied to the FDL or FHL muscle at rest, with the treadmill off and in the absence of movement, and during periods of walking (e.g. Fig. 1). The stretches were uncorrelated with the step cycle. A larger stretch amplitude (i.e. 1 mm) was used with group II afferents since they are less sensitive to stretch. After recovery from anaesthesia the cat walked with three legs on a treadmill while the fourth, denervated limb was supported in a fixed position. Locomotion occurred spontaneously or in response to electrical stimulation (16 Hz, 0.5 ms width, 100–150 μA) of the mesencephalic locomotor region (MLR: co-ordinates P2, L4, H6; Shik et al. 1966). Similar afferent responses were observed under both conditions. Over a period of time, stretches occurred at all phases of the locomotor cycle. Data were amplified and monitored on oscilloscopes. Neural and EMG activity, length, tension and stretch markers were recorded on VHS tape using a Vetter 4000A PCM recording unit (Wintron Technologies, Howard, PA, USA) for later analysis, as described below.

Figure 1.

Muscle and afferent activity during locomotion

EMGs from FDL and FHL were rectified and low-pass filtered. FHL showed a single burst of EMG activity (extension phase) per step cycle while FDL EMG had two components: early low level activity related to extension and a later more prominent burst around the end of this phase. Manually placed markers (asterisks) at the beginning of FHL EMG bursts denote the beginning and end of step cycles. Stretch markers (arrows) are also indicated that correlate with sinusoidal length changes (0.25 mm amplitude, 5 Hz; lengthening phase is upwards) applied to the FHL muscle, which occurred independently of the phase of the step cycle. Muscle force recordings indicate substantial active tension in FHL while FDL was recruited only weakly. The instantaneous firing rate of a FHL Ia afferent is shown at the bottom. The afferent's discharge represents the response to both locomotion and stretch. Periods of flat instantaneous frequency (at high values) reflect the 1 kHz digitisation rate used in data analysis (see Methods). The distance between the tick marks on the y-axis represents 15 N (FHL force), 1 N (FDL force) and 400 impulses s−1 (unit). EMG units are arbitrary.

At the end of experiments, the death of the animals was ensured by an intravenous administration of an overdose of barbiturate.


Data were replayed from the tape recorder. Afferent action potentials were converted into standard pulses. Data were digitally sampled using AxoTape software (Axon Instruments, Union City CA, USA) at a frequency of 1 kHz and stored on disk. The sampled data were then analysed using specially written programs under MATLAB (MathWorks, Natick, MA, USA). The EMGs were rectified and low-pass filtered (30 Hz). Length and tension signals were also low-pass filtered (30 Hz). Markers were placed manually at the beginning and end of step cycles corresponding to the time of onset of FHL EMG activity during regular periods of walking. Figure 1 illustrates data processing with EMG recordings during walking, together with corresponding muscle tensions, and step cycle markers. The discharge pattern of a Ia afferent from FHL is also shown during sinusoidal length changes of the parent muscle that were used to generate stretch markers. Various types of analysis were carried out as follows. Cycle histograms (bin width, 1 ms) were computed, using the stretch or step markers, for the afferent pulse trains, in addition to averaging muscle length, tension and EMG. Histogram duration was adjusted to correspond to either one stretch cycle (e.g. Fig. 2A) or one walking cycle (e.g. Fig. 2B). Locomotor averages that were normalised to each step cycle were also generated (e.g. Fig. 3). All histograms were low-pass filtered (30 Hz).

Figure 2.

Cycle histograms related to the stretch cycle and the step cycle

A, the response of a muscle spindle Ia afferent to sinusoidal stretch during locomotion is displayed as a smooth histogram, which has been fitted with a sine curve (dotted line), to determine stretch modulation. Characteristically, the afferent's response was phase advanced relative to stretch indicating length and velocity sensitivity at the applied frequency of stretch. B, smooth histogram of Ia afferent impulse rate throughout the step cycle is shown together with average EMG and force from the parent muscle. Data represent the average of 126 cycles. A and B were generated from the same FHL spindle Ia afferent with stretch and step markers, respectively, as shown in Fig. 1 during the same period of walking in which sinusoidal length changes (0.25 mm amplitude, 4 Hz) were applied throughout. The stretch markers occurred independently of the phase of the step cycle. EMG units are arbitrary.

Figure 3.

FHL muscle spindle afferent discharge and modulation during locomotion

Mean firing rate and stretch modulation of a Ia (A) and a group II (B) muscle spindle afferent to sinusoidal length changes (4 Hz, amplitude: A, 0.25 mm. B, 1 mm) at different phases of the step cycle. EMG and force changes show the timing of activity in the homonymous muscle. For both types of afferent the discharge rate was increased and modulation reduced throughout the step cycle, compared to resting values (dotted lines). In addition the Ia afferent showed an increasing trend in modulation during the extension phase. Data represent the average of 66 (A) and 61 (B) step cycles. Mean cycle durations were 1075 ms and 992 ms. The y-axes units are impulses s−1 (mean rate and modulation), arbitrary (EMG) and N (force). The afferents were from different experiments. In both cases locomotion occurred spontaneously.

The interaction between stretch and walking on muscle afferent discharge was analysed according to the method previously described in Taylor et al. (1985) and will be summarised here. The total step cycle was divided into sixteen phases and the average response of the muscle afferent to stretches that began in each phase was determined separately from stretch cycle histograms. The mean rate during each phase was calculated from each stretch cycle histogram. Afferent discharge will be affected by both the stretch and locomotion. As a first approximation we assumed that responses to stretch and locomotion summed linearly (Taylor et al. 1985). The afferent's rate was averaged over the step cycle (e.g. Fig. 2B) from the same period of locomotion during which stretch was applied. To obtain the modulation in afferent rate due to stretch, the afferent's average rate over the corresponding step cycle phase (Fig. 2B) was subtracted from each stretch cycle histogram. Each of the stretch cycle histograms was then fitted with a sine curve (least mean squares deviation) and the modulation in rate determined. A plot of mean rate and modulation of afferent discharge versus phase of the step cycle could then be generated. Values were plotted at the mid-point of the stretch cycle in graphs such as Fig. 3. Sine curve fits were also fitted in characterising variations in mean afferent rate and modulation. The sine curve had the form:

y/A+B sin x+C cos x,

which is equivalent to:

y/A+ Mag cos (x - Pha),

where A is the mean value, Mag is the modulation (half peak-to-peak) and Pha is the point at which the curve reaches its peak. The statistical significance of differences between mean values was analysed by Student's two-tailed t test. In all statistical tests, P < 0.05 was accepted as being significant. Results are expressed as means ±s.e.m., unless otherwise stated.


FHL afferents

Figure 2B illustrates an example of the discharge of an FHL Ia afferent during the step cycle. The firing was clearly modulated with changes related to muscle force. A reduction occurred as tension rose, while firing increased on the falling phase and peaked around its maximum rate of change. Because no γ patterns have been recorded that could account for this behaviour (Murphy, 2002) it is likely that mechanical factors, due to unloading and relengthening of the spindle during extrafusal muscle contraction, were responsible. Similar mechanical effects were observed in most Ia and group II afferents, but were generally less marked in the latter case. In addition, a few units (two Ia, one group II) showed an increasing trend in rate, prior to EMG onset, when active muscle tension was absent. This feature may be due to the action of tonically active γ-efferents which fire throughout the step cycle and sometimes peak before FHL α activity begins (Murphy, 2002).

During locomotion the stretch stimulus could occur at any time in the step cycle. Stretch responses at 16 different phases were analysed separately (for details, see Methods) and the results for an Ia afferent are illustrated in Fig. 3A. The unit had an increased rate, but decreased modulation, throughout the step cycle compared to rest (dotted lines). These features are good evidence for raised static γ drive at all phases (Taylor et al. 1985); a conclusion supported by locomotor data from group II afferents, which are almost exclusively influenced by static γ-axons (Matthews, 1981). Thus the group II unit in Fig. 3B also showed a sustained increase in rate and reduction in modulation. The modulation of all FHL afferents in the present sample was reduced throughout the step cycle. Most Ia (10 of 18) and group II (7 of 10) afferents also had raised firing levels at all locomotor phases. For the remainder, with one exception, discharge rates were decreased during the extension phase (duration of FHL EMG burst) but increased in flexion, compared to rest. The enhancement was probably due to tonically active γ-efferents since only this category had maintained firing during flexion (Murphy, 2002). As these neurones had elevated discharge throughout the step cycle, it is likely that mechanical unloading was responsible for the lack of raised afferent firing in extension. Tonically active γ-efferents had a weaker phasic component and generally peaked during the extension phase. It is possible that extrafusal effects obscured signs of modulated static fusimotor drive at this time. Thus almost all the data suggest raised static γ drive at all phases of the locomotor cycle. Exceptionally, the firing of one Ia afferent was increased during extension but decreased in flexion, compared to rest.

The values of rate and modulation at mid-extension and mid-flexion were taken as estimates of the magnitude of effects of γ drive in these phases, with their mean representing the step cycle as a whole (Table 1). During locomotor activity the mean discharge rates of FHL Ia and group II afferents were significantly increased (P < 0.05, Table 1) in mid-flexion and mid-extension, compared to rest. The corresponding values for modulation were significantly reduced (P < 0.01) at both mid-flexion and mid-extension. Over the step cycle the mean discharge rate of Ia and group II afferents increased by 75 % and 111 %, respectively, accompanied by reductions in modulation of 63 % and 77 %.

Table 1.  Summary of FHL and FDL muscle spindle afferent discharge rates and stretch modulation at rest and during locomotion
  IaGroup II
  1. Firing during locomotion is shown at mid-extension and mid-flexion. Discharge rates and modulation to sinusoidal stretch (4–5 HZ, amplitude: 0.25 mm for Ia, and I mm for group II, afferents) are in units of impulses s−1 (means ±s.e.m.). n/ number of units. Mean 2 and 3 denotes the mean of extension plus flexion.

 1. Rest32 ± 335 ± 328 ± 522 ± 2
 2. Extension42 ± 414 ± 248 ± 115 ± 1
 3. Flexion71 ± 712 ± 169 ± 86 ± 2
 4. Mean 2 and 356 ± 413 ± 159 ± 95 ± 1
 1. Rest21 ± 226 ± 333 ± 523 ± 3
 2. Extension61 ± 89 ± 177 ± 1310 ± 1
 3. Flexion45 ± 610 ± 171 ± 1112 ± 2
 4. Mean 2 and 353 ± 610 ± 174 ± 811 ± 2

Although the modulation of the Ia afferent in Fig. 3A was decreased throughout the step cycle, compared to rest, there was an increasing trend during the extension phase. The trend continued during a period of relatively steady discharge rate and muscle force (20-40 % step cycle, Fig. 3A), and is consistent with phasic dynamic γ drive related to extension. In 13 of the 18 Ia afferents, a clear variation of stretch modulation was observed during extension. Peak modulation occurred, on average, 53 ± 4 % of the step cycle after EMG onset. In order to further characterise this variation, a sine curve was fitted (see Methods). The units that were classed by eye as having a trend during extension had peak modulation during this phase (39-66 % step cycle, 12 units) or in early flexion (67 % step cycle, 1 unit), with variations in modulation between 2.9 and 8.9 impulses s−1 that accounted for 48–95 % (r2) of the variance. The modulation of two of the remaining units peaked in flexion (74 % step cycle, r2/ 36–65 %) while the others had variations that were not significant (1 unit, r2= 16 %, P > 0.1) or ≥ 1.5 impulses s−1 (2 units, r2= 42–45 %). The lack of signs of phasic dynamic γ drive during extension in these afferents may reflect an absence of contacts with a bag1 intrafusal muscle fibre, as occurs in some spindles in other muscles (Price & Dutia, 1989; Taylor et al. 1992). Thus most FHL Ia afferents probably received tonic static and phasic dynamic γ drive during locomotor activity. Group II afferents showed little sign of periodically increased modulation during extension (e.g. Fig. 3B). Two units had no significant variation (r2= 3–19 %, P > 0.05). For most units modulation peaked in flexion (82 ± 6 %, n= 4; r2= 32–75 %) or variations were ≥ 1.1 impulses s−1 (2 units, r2= 30–31 %). The modulation of two group II afferents peaked during extension with variations (1.9 and 2.4 impulses s−1) that accounted for 54–73 % of the variance. Overall, on the basis of both firing rate and modulation, group II afferents lacked indications of phasic static γ drive in extension.

FDL afferents

Like FHL, Ia (12 of 14) and group II (6 of 7) afferents from FDL typically showed an increase in rate and a decrease in modulation throughout the step cycle, compared to rest (e.g. Fig. 4A and B), suggesting elevated static γ drive at all phases. These changes, which were significant (P < 0.02) both in mid-extension and mid-flexion, represent average rises of 152 % and 124 % in Ia and group II afferent discharge rate, respectively, over the step cycle (Table 1). Corresponding reductions of 62 % and 52 % occurred in modulation.

Figure 4.

FDL spindle afferent discharge and modulation during locomotion

Ia (A) and group II (B) afferent firing was increased and modulation to stretch (4 Hz, amplitude: A, 0.25 mm. B, 1 mm) reduced throughout the step cycle, compared to rest (dotted lines). Both types of afferent showed rhythmic variation in discharge rate, with increased activity and peak firing during the extension phase (duration of FHL EMG burst). Data represent the average of 98 (A) and 71 (B) step cycles. Mean cycle durations were 1085 ms and 1097 ms, respectively. Format is the same as that of Fig. 3. MLR stimulation was used to induce locomotor activity. The afferents were from the same experiment.

An additional feature of afferent behaviour concerned their patterns of discharge. In about half the cases (seven Ia and three group II afferents) clear cyclic variations were apparent, with increased activity and peak firing (24 ± 2 % step cycle after EMG onset) during the extension phase (Fig. 4A and B). A sine curve gave good fits (r2≥ 0.84) to these units and maximum rates occurred between 16–35 % of the step cycle (i.e. in extension). Because discharge rate did not appear to be related to changes in muscle tension (e.g. Fig. 4), it is unlikely that mechanical unloading and relengthening of spindles due to extrafusal contraction were responsible for the cyclic firing pattern. As both Ia and group II afferents were involved, phasic static γ activity is implicated in these cyclic patterns of discharge. With one exception (Ia unit), afferents with this pattern also had enhanced firing rates and reduced modulation throughout the step cycle.

Interestingly, cyclic variations in Ia and group II afferent discharge were generally not accompanied by obvious changes in modulation, possibly due to saturation (cf. Goodwin et al. 1975). Thus, in Fig. 4A, Ia afferent modulation was reduced to a relatively steady level during the step cycle even though firing rate varied phasically under fusimotor drive. Similarly, the modulation of the group II afferent in Fig. 4B showed relatively little change despite receiving powerful phasic static γ drive during the extension phase. The discharge rates of all such cyclic units (Ia and group II) were higher at mid-extension than mid-flexion. On average their firing rate was approximately doubled in mid-extension (Ia: 77.7 ± 9.7 impulses s−1, n= 7; group II: 105.7 ± 18.4 impulses s−1, n= 3) compared to mid-flexion (Ia: 43 ± 8.6 impulses s−1, n= 7; group II: 58.7 ± 4.4 impulses s−1, n= 3) with a significant difference (P < 0.05) for Ia afferents. In contrast, corresponding mean values of modulation were similar (Ia: 10.5 versus 11.7 impulses s−1; group II: 9.4 versus 12.9 impulses s−1).

In a previous report, using the same preparation as in the present study (i.e. premammillary decerebrate cat), the pattern of FDL α activity was variable and two components were distinguished (Murphy, 2002). One was related to extension while the other comprised a brief burst around the end of this phase (cf. Fig. 4A). Their relative sizes could vary and frequently overlapped in time giving the appearance of a single period of activity approximately coactivated with ankle extensors. These features were also observed in the present study. In addition, muscle tension records indicated that FDL was weakly activated (maximum, 0.7 N) during locomotor activity (cf. O'Donovan et al. 1982; Murphy, 2002), while FHL generated considerable force (maximum, 13 N). Despite variations in FDL activity, no associated differences were observed in the pattern of afferent firing or modulation. Figure 5 shows data from the Ia afferent illustrated in the preceding figure (late FDL burst present) during an earlier period of locomotor activity when no late α burst was apparent in FDL. The patterns of modulation and discharge rate were similar in the two cases, suggesting a lack of γ activity associated with the late burst in FDL, a conclusion that is also supported by direct γ recordings (Murphy, 2002). The earlier period of locomotor activity was less vigorous as indicated by the lower force level in FDL (compare Fig. 5 and Fig. 4A). Indeed there was actually a negative trend in the FDL force record during the first half of the step cycle, superimposed upon the effect of sinusoidal stretch, despite EMG activity in the parent muscle (Fig. 5). This trend was related to the rising phase of FHL muscle tension and presumably reflected mechanical unloading of FDL by the more strongly activated FHL which lies in parallel.

Figure 5.

FDL Ia afferent firing and modulation during locomotion in the absence of a late burst in homonymous α activity

The patterns of discharge rate and modulation to stretch (4 Hz, 0.25 mm amplitude) are similar to Figure 4A (same unit) despite the absence of a late peak in FDL EMG and tension. FHL EMG and tension are shown for comparison. Data represent the average of 49 step cycles having a mean duration of 1251 ms. Locomotion occurred spontaneously. The y-axes units are impulses s−1 (mean rate and modulation), arbitrary (EMG) and N (force).

Figure 6 illustrates a typical example of an FDL unit without extension-related cyclic firing. The Ia afferent showed an increasing trend in rate, before EMG onset, when active muscle tension was low or absent in both FDL and FHL. Firing peaked around the beginning of EMG activity and subsequently declined. However, the possibility that phasic γ drive accompanied the extension phase cannot be excluded since spindle unloading due to contraction of FDL and/or FHL may have offset its effects. Eight further units had a similar discharge pattern. Sine curve fits gave peak firing at 92–12 % of the step cycle (i.e. around extension onset, n/ 9) and accounted for 51–93 % of the variance. Two other units had maximum discharge near the end of extension (38 and 52 % step cycle) with r2 values of 0.63 and 0.6, respectively. Units lacking extension-related cyclic activity had lower, or similar, discharge rates in mid-extension (Ia: 44.8 ± 8.1 impulses s−1, n= 7; group II: 55.2 ± 9.5 impulses s−1, n= 4) compared to mid-flexion (Ia: 46.7 ± 8.6 impulses s−1, n= 7; group II: 79.7 ± 18.5 impulses s−1, n= 4). Corresponding mean values for modulation were as follows: Ia, 10.4 versus 11.6 impulses s−1; group II, 8 versus 7.8 impulses s−1. Increasing trends in firing that preceded EMG onset were frequently observed in FDL (seven Ia and six group II afferents) and could occur in units with an extension-related cyclic discharge pattern. They probably represent the activity of tonically active γ-efferents that generally peak before EMG commences (Murphy, 2002).

Figure 6.

FDL Ia afferent without extension-related rhythmic firing

Firing increased during the flexion phase, peaked around EMG onset and subsequently declined. FHL EMG and tension are shown for comparison. Data represent the average of 53 step cycles with a mean duration of 1216 ms. Format is the same as that of Fig. 5.


Nature and patterns of fusimotor drive

In a previous study (Murphy, 2002) the activity of FDL and FHL γ-efferents was recorded directly in a similar preparation (i.e. premammillary decerebrate cat) to that used here. Two basic patterns were distinguished in both muscles. ‘Tonic’ units fired throughout the step cycle and had less modulation, but higher minimum rates, than ‘phasic’ units which were mainly recruited with ankle extensor EMG activity. In both muscles tonic units had low resting rates (0-8 impulses s−1) while phasic units showed a wider range (0-50 impulses s−1). What are the implications of the present spindle recordings for the nature of these fusimotor patterns? In FHL and FDL tonically active γ-efferents showed a marked increase in mean rate between the resting and locomotor states. Further, only the tonic category usually discharged in the extension and flexion phases. Thus, the increased firing rate and decreased modulation of FDL and FHL spindles during locomotion is most simply explained if most tonic γ-efferents are static in nature.

The spindle afferent data suggest differential phasic fusimotor action during the extension phase in FDL and FHL. Signs of phasic dynamic γ drive occurred only in FHL. While phasic static γ drive was indicated for FDL, modulated static activity could not be excluded with FHL. Regarding FHL, phasic γ-efferents were recruited with ankle extensor EMG activity (Murphy, 2002). The modulation of FHL tonic γ-efferents was lower and peaked shortly after EMG onset (mean, 2 % step cycle). Such activity is unlikely to be responsible for the cyclic variation in modulation of FHL Ia afferents that peaked 53 % (mean) through the step cycle (see Taylor et al. 1985). Since Ia afferents are typically influenced by 1–2 dynamic γ-efferents (Matthews, 1981), these observations suggest that most FHL dynamic γ-motoneurones are phasically recruited during locomotion.

In FDL, phasic γ-efferents were recruited with extension while the tonic category had weak modulation (mean, 10 impulses s−1) and generally peaked before (mean, 9 % step cycle) extensor EMG onset (Murphy, 2002). The marked increase in discharge of half the FDL Ia and group II afferents during extension, peaking 24 % (mean) after EMG onset, therefore suggests that a component of the phasic γ population is static in nature. As dynamic γ-axons excite Ia afferents (but not group II), it is possible that this type is also phasically active in FDL. However, any such γ activity was never dominant, as indicated by the lack of accompanying increases in afferent modulation.

Because about 40 % of FDL afferents (9 of 21, e.g. Fig. 4B) had enhanced discharge rate and reduced modulation throughout the step cycle (due to tonic static γ-efferents), and showed increased firing during extension (involving phasic static γ-efferents), it follows that a given spindle commonly received distinct patterns of static γ discharge through separate γ-efferents (i.e. tonic and phasic). The functional significance of this dichotomy, which implies specificity of locomotor control within the static γ system, is unknown but may be related to the suggestion that there are two types of static γ-axon that differ in the distribution of their effects on bag2 and chain intrafusal fibres (Boyd, 1986; see also, Taylor et al. 1998). Alternatively there may be a need to channel distinct patterns of γ drive via different, but functionally identical efferents. Evidence of two locomotor patterns of static γ discharge has also been reported recently on the basis of recordings from the medial gastrocnemius nerve in decerebrate cats (Taylor et al. 2000b). However, it should be noted that in the original recordings of fusimotor activity from this nerve, static γ sub-groups were not distinguished (Murphy et al. 1984). The reason for the apparent disparity between these studies is currently unclear.

Very little is known about the discharge patterns of identified γ-efferents during locomotion. The data involve hind limb muscles of decerebrate cat preparations (for review see Murphy & Martin, 1993): triceps surae (ankle extensor), medial sartorius (hip/knee flexor) and tibialis anterior (ankle flexor). The tonic static and phasic (α-linked) dynamic γ-drives described here for FHL are very similar to those of triceps surae. Functionally this is appropriate since the muscles share a marked extensor synergy and contribute to anti-gravity support. Tibialis anterior and medial sartorius are active during flexion, and have a strong phasic (α-linked) static γ drive in common. FDL has both a flexor (late stance) and extensor α synergy (Fleshman et al. 1984; Trank & Smith, 1996) during which the muscle shortens and lengthens, respectively (O'Donovan et al. 1982). However, its γ drive does not match other muscles with either synergy since the FDL flexor α burst was not linked to γ activation, and its extensor α drive was accompanied by phasic static γ activation. It would thus appear that α synergy does not necessarily correlate with fusimotor usage. Further, phasic static γ activity in FDL during extension, as predicted from the present results, would not be related to muscle shortening, in contrast to other muscles during rhythmic movements (for review see Murphy & Martin, 1993).

The rules that govern the γ system during movement have yet to be fully established. α/γ linkage appears to be the norm in man while α/γ independence is common in the cat (for review see Prochazka, 1996). Different experimental conditions may, however, underlie the apparent discrepancy. In both cases, a major unresolved problem concerns the γ drive to different muscles during the same behaviour. Does it vary? If so, what are the governing rules and functional significance? The present results add to the growing body of evidence, from reduced cat preparations, of muscle-specific γ drive during locomotion (Perret & Berthoz, 1973; Cabelguen, 1981; Bessou et al. 1990; Murphy & Hammond, 1993; Murphy, 2000; Taylor et al. 2000a; Murphy, 2002). Various factors may influence the strategy of γ usage for a particular muscle including its kinematic behaviour (i.e. motion) and activation state (see later).

Functional implications

There are few spindle afferent recordings from FDL or FHL in the intact cat with which to compare the present data (Prochazka et al. 1976; Loeb & Duysens, 1979; Prochazka & Gorassini, 1998). Nevertheless, on the reasonable assumption that the γ locomotor discharge characteristics of the decerebrate cat are representative of the basic patterns during normal movement (Murphy & Martin, 1993; Taylor et al. 2000a; Murphy, 2002), certain functional advantages may arise, depending on the setting of central pathways, from the suggested strategies of γ usage. In FDL and FHL the proposed tonic static γ drive would increase the rate and reduce the sensitivity (i.e. stretch modulation) of Ia and group II afferents during locomotion. Direct recordings of FDL muscle length indicate that marked changes occur over the step cycle, but with little active tension (O'Donovan et al. 1982). It has been proposed that the fusimotor system optimises the sensory information obtained from the spindle to the expected kinematic conditions (Loeb, 1984; Scott & Loeb, 1994). In this context, tonic static γ drive in FDL would be functionally useful in preventing receptor saturation at both ends of the dynamic range. Direct information concerning the normal kinematics of FHL is currently unavailable and a similarity with FDL cannot be assumed because of substantial differences in their moment arms at the ankle (Young et al. 1993).

Concerning motor function, since FHL and triceps surae have similar γ drives and share a prominent anti-gravity function during locomotion, it seems reasonable to suggest a common fusimotor role in contributing to the adjustment of stretch reflex gain to a level appropriate for the ongoing movement (Taylor et al. 1985; Bennett et al. 1996a,b). In FDL, tonic static γ drive may be related to the activation state (low, brief) of the parent muscle. A low Ia afferent sensitivity would minimise stretch reflex responses to peripheral perturbations and support weak (but precise) central α activation.

The various proposals concerning the motor role of γ drive are dependent upon the operative status of central pathways. Experiments involving electrical stimulation of group I muscle afferents in the decerebrate cat have demonstrated homonymous mono- and disynaptic excitation of FDL and FHL α-motoneurones during fictive locomotion (Degtyarenko et al. 1998). The disynaptic pathway was facilitated during extension, as occurs in hindlimb extensors (McCrea et al. 1995). These observations offer a range of potential reflex control strategies involving γ drive and muscle spindle Ia afferents, and tendon organ Ib afferents (for review see Pearson, 1995), including the preceding suggestions. Further, since heteronymous monosynaptic Ia connections exist between FHL and FDL (Fleshman et al. 1984), the intriguing possibility arises that the phasic (α-linked) γ drive of FDL (or FHL) may influence the spindle reflex regulation of its anatomical synergist, even though it is functionally distinct.

Recordings of muscle spindle afferents from the intact cat during natural movements have indicated that fusimotor drive is task- and context-dependent; observations that led to the fusimotor set hypothesis (Prochazka, 1983, 1996; Prochazka & Hulliger, 1983). One aspect of this view is that ‘proprioceptive sensitivity is adjusted at its source by the CNS according to the overall sensorimotor requirements predicted for upcoming movements’ (Prochazka, 1989, p. 289). To date, evidence consistent with muscle-specific fusimotor drive, based on afferent recordings in intact animals, is sporadic (e.g. Loeb & Duysens, 1979). A full appreciation of this feature of fusimotor control may require the development of new techniques that permit direct recordings from identified and classified γ-motoneurones during natural behaviour. If muscle-specific γ drive is prevalent under such conditions then it is compatible with the fusimotor set hypothesis.


The authors thank R. Gramlich for excellent technical assistance and Dr J. Misiaszek for help with preparatory surgery. A travel grant from the Royal Society to Dr P. Murphy is gratefully acknowledged.