JOURNALOF NEUROPHYSIOLOGY Vol. 54. No. 3. September 1985. Prinled in U.S.A.

Activity of Spindle Merents From Cat Anterior Muscles. I. Identification and Patterns During Nornid Locomotion

G. E. LOEB, J. A. HOFFER, AND C. A. PRATT Laboratory of Neural Control, IRP, National Institute of Neurological and Communicative Disorders and Stroke, National Institutes of Health, Bethesda, Maryland 20205

SUMMARY AND CONCLUSIONS 6. Activity from spindle secondary endings 1. The naturally occumng activity patterns was generally lower, more regular, and less ve- of anterior thigh muscle spindle afferents were locity dependent than that from primary end- recorded during unrestrained treadmill loco- ings, consistent with their lack of input from motion by means of floating microelectrodes the dynamic fusimotor apparatus. chronically implanted in the fifth lumbar dor- 7. The activity of all spindle afferents stud- sal root ganglion. ied was similarly well modulated during ex- 2. Conduction velocity of units from pri- trafusal activity of the parent muscles, regard- mary and secondary endings was determined less of the kinematic conditions of muscle by spike-triggered averaging of the signals from length and velocity during which this muscle a chronically implanted cuff. work occurred. This suggests that the fusi- 3. Activity from extensor muscle motor apparatus is well orchestrated to regu- spindles generally occurred during periods of late the static and dynamic sensitivity of pri- muscle lengthening, but was often greater for mary spindle afferents at levels appropriate to small stretches when the muscle was active the anticipated motion. (during stance phase of walking) than for larger INTRODUCTION stretches when the muscle was passive (swing phase), indicating fusimotor enhancement of By virtue of their numbers, speed, and sen- spindle sensitivity in phase with extrafusal sitivity, muscle spindle afferents seem well muscle recruitment. suited to providing much of the information 4. Activity from spindles in biarticular available to the spinal cord regarding the po- muscles acting across the knee and was sition and motion of the limb segments. Their more variable and complex than that seen in strong and widespread synaptic terminations the pure knee extensors, and frequently in- directly on motoneurons and on interneurons cluded activity during rapid muscle shortening subserving local motor control circuits, as well (swing phase) indicative of strong static fusi- as their contribution to ascending pathways, motor input. support this presumption further. However, 5. Changes in speed of gait caused changes the nature of the information that they provide in the range and velocity of muscle length ex- is difficult to predict from their receptor prop- cursions monitored by chronically implanted erties because these properties are so highly length gauges, but such changes were accom- dependent on the mechanical activity of the panied by only modest changes in spindle af- intrafusal muscle fibers, which are under the ferent activity, suggesting concurrent and continuous and rapidly modulatible control compensatory changes in fusimotor influence of gamma and beta motoneurons (for reviews on spindles. see Refs. 7, 29). 550 LOEB, HOFFER, AND PRATT

In this and the accompanying two papers, ity and nonlinear interaction of the various we report the results of chronically recording fusimotor influences on spindles preclude the activity of single muscle spindle afferents quantitative assessment of fusimotor activity during unrestrained, normal behavior. We from these records, they do provide a general have examined the effects of various behaviors indication of the degree to which normal mo- on the afferent activity of different spindles in tor behavior is accompanied by fusimotor the same muscle and across different muscles programs that are specific to each muscle and and have contrasted these activity patterns the task of the moment. Furthermore, because with the sensitivity of these afferents to similar of the distinctive effects of gamma dynamic length changes applied to relaxed muscles in and gamma static fusimotor input on spindle the anesthetized state. Although the complex- afferent5 during states such as rapid muscle

FIG. I. A: innervation of the anterior thigh muscles of the cat and the relationship to the instrumentation implanted for these studies. DRG, array of up to 12 individual metal microelectrodes implanted in the 5th lumbar dorsal root ganglion; IT, tripolar electrode configuration in the proximal half of the cuff electrode; FD, tripolar electrode configuration in the distal half of the femoral nerve CURSaph. n. and Hamst. n., bipolar nerve cuffs used for stimulation experiments reported elsewhere (25); SA-a, bipolar EMG patch electrode on anterior part of sartorius muscle; SA-m, bipolar patch electrode on medial sartorius; RF, multipolar EMG spiral electrode in ; VM, spiral electrode in muscle; VI, spiral electrode in vastus intermedius muscle; VL, spiral electrode in ; LV,length gauge across vastus muscles (knee joint only); LR,length gauge across rectus femoris and anterior sartorius muscles (knee and hip joints); Fp, strain gauge mounted on patellar ligament. SPINDLE ACTIVITY 551 shortening, it is occasionally possible to draw This paper describes the functional anatomy specific inferences about the nature and and kinematic and electromyographic patterns strength of fusimotor influence. of the anterior thigh muscles during normal Afferent unit recordings were obtained from locomotion, together with the simultaneously cell bodies or axons located in the fifth lumbar recorded activity of identified single spindle dorsal root ganglion (L5 DRG). About half of afferents. Two companion papers describe our the afferents in this ganglion derive their input attempts to dissociate intra- and extrafusal ac- from receptors in the median articular nerve tivity by fusimotor blockade (23) and by reflex- of the knee, the cutaneous , generating stimuli (25). Preliminary reports and the motor innervating the sartorius regarding the activity of 8 of the 23 units re- and quadriceps femoris muscles (see Fig. 1). ported here have been published previously All of these nerves together comprise the entire (10, 18, 24); none of the activity patterns femoral nerve as it passes through the inguinal shown here include those units, but conclu- canal, where its long, straight, and unbranched sions and summaries are based on the entire course permits the chronic implantation of a population. multilead nerve cuff. By using the method of spike-triggered averaging of the neurogram MATERIALS AND METHODS recorded at two points along the nerve, a uni- tary potential recorded in the L5 DRG can be Ten adult male cats ( Felis domesticus) were im- unequivocally associated with an afferent fiber planted with all or most of the various devices in- having a particular conduction velocity and dicated in Fig. 1, with aseptic surgical techniques innervating one of the above noted receptor- and general anesthesia consisting of intraperitoneal bearing structures. Because these structures are pentobarbital (45 mg/kg) followed by maintenance doses of intravenous pentobarbital. Surgery was all large, superficial, and relatively simple me- usually performed in two stages, with the peripheral chanically, manual manipulation and the ap- transducers of length and force and the nerve and plication of vibrating probes can be used to muscle electrodes implanted one week prior to im- further localize and characterize these afferents plantation of the recording microelectrodes in the with little ambiguity. DRG. All animals were walking normally within

HG. 1. B: schematic representation of the typical walking step cycle showing the stance phase (E2-3,heavy line at bottom) and 2 complete swing phases divided into flexion phase (F) and first extension phase (El). The length and EMG records indicate the most commonly seen pattern under technically ideal conditions. Actual records tended to have minor variations as a result of variability of gait among steps and across animals and occasional hardware problems such as inadequately tensioned length gauges or mispositioned EMG electrodes. 552 LOEB, HOFFER, AND PRATT several days after each surgery when the effects of desired length using scissors. Each gold lead, in turn, anesthesia had worn off. Unit records were usually was welded to a multistranded stainless steel lead obtained in the second to fifth postoperative weeks. anchored within silicone rubber tubes mounted on Both the implanted devices and the external con- the dorsal spinous processes, to provide a flexible nector were well tolerated throughout the course of and strong percutaneous lead to the connector the experiment, with no apparent discomfort or ir- mounted outside on the back of the animal. The ritation either during recording sessions or between welds were reinforced with epoxy, and the entire them, although some animals favored the operated assembly overcoated with Parylene insulation, the leg for the first few postoperative days (see Fig. 5). deposition of which is detailed elsewhere (2 1). The Figure 2 shows a detail of the microelectrode im- L5 DRG was exposed through the usual midline plantation and fixation technique together with the dorsal dissection of soft tissues followed by a small general strategy for maintaining a stable, long-term burr hole (-3 X 5 mm) through the overlying bony connection to the numerous implanted devices. The lamina. These electrodes were inserted by hand into microelectrodes consisted of 50-pm diameter, plat- the DRG through small scalpel slits, with fine for- inum-20% iridium shafts with Pyre Tri-ML poly- ceps with tips ground to receive the reinforcing imide insulation (California Fine Wire Co.) cross- epoxy ball. Typical electrode impedances were 100- welded to 25-pm diameter gold lead-out wires for 300 kQ at I kHz; impedance was measured daily flexibility. The recording surface was formed at sur- prior to recording and proved to be a sensitive in- gery by simply cutting the shaft obliquely to the dicator of the loss of recording sources due to

FIG. 2. Detail of microelectrode, nerve cuff, and back-pack connector fixation. The silicone rubber tube array (TA) mounted on the L, and L6 dorsal spines carries the joint between the gold leads of the dorsal root ganglion (DRG) electrodes and the stranded, stainless steel percutaneous leads (L) that are soldered to the printed circuit boafd (PCB) mounted as a back pack via anchoring sutures (AS) through the L4 and L, dorsal spines. The femoral nerve cuff (FNC) carries five circumferential contacts whose percutaneous leads (L) terminate similarly on PCB along with an intraluminal catheter (IC). For recordings, multipin connector C mates with a connecting ribbon cable and 12-channel preamplifier; between sessions, a light-weight aluminum cap protects the percutaneous leads and joints from being snagged. SPINDLE ACTIVITY 553 breakage of leads or insulation. Typically, a total should always be correct, but the absolute value of 12 such electrodes were implanted in the LSspinal may not be similarly calibrated over the entire ex- root structures, including variable numbers of sim- cursion range of muscles with long lever arms such ilar ventral root electrodes whose recordings are de- as sartorius. Velocity calibrations were taken from scribed elsewhere ( 13). the midrange length excursions. Also, the method The external connector assembly consisted of a of electronic differentiation to obtain velocity had convexly curved sheet of Dacron reinforced silicone to be adjusted to compensate for high-frequency rubber (Dow Corning Silastic #501-7) on which a noise in some length recordings, producing slight 3 X 10 cm printed circuit board carrying a 40-pin phase lags in some velocity traces (e.g., Fig. 4). ribbon cable connector was affixed (3M Scotchflex The strain gauge on the patellar ligament em- #3432), termination pads for soldering the percu- ployed a two-arm balanced-bridge configuration of taneous leads, and stainless steel nipples for receiv- semiconductor elements (BLH #SPB 1-20-35-U 1) ing the implanted catheters and their caps. The as- epoxied to an E-shaped stainless steel substrate spe- sembly was anchored to the animal's back by snugly cifically designed for this broad, triangular shaped tied #5 Ethibond sutures that had been passed ligament (30). Frequency response was flat from through the skin and subcutaneous tissue and DC to 100 Hz; calibrations are not given because through holes drilled in the dorsal spinous processes they drift over the implantation time and can only of the L4 and L7 vertebrae. This floating mount was be obtained accurately in situ in a terminal exper- tolerated without imtation or infection for many iment (26). These gauges tend to be quite linear, months and accepted the load of the preamplifier and the records are intended for qualitative com- array and connector cable without interfering with parisons of force output and its temporal waveform; the animal's normal activities. gains and offsets are constant in any single figure. All of the other implanted devices have been de- The femoral nerve cuff consisted of a 2- to 3-cm- scribed previously. The two length gauges were long, longitudinally slit section of silicone rubber made from silicone rubber tubing (0.044" ID by tubing with a 2.8-mm ID (Sil-Med Corp.), equipped 0.065" OD, Sil-Med Corp.), filled with hypertonic with five equally spaced circumferential contacts saline (45 g/liter NaCl in water with blue vegetable made from stranded platinum-10% iridium wire coloring), and equipped with stranded stainless steel (Medwire lOIR9149T). It was slipped over the fem- electrodes and leads in each end (26). They were oral nerve and tied closed with several circumfer- anchored proximally to bone screws and distally by ential sutures. Starting with the third cat in the series a criss-cross tendon suture just above the reported here, the nerve cuffs were equipped with (Fig. I). Length changes were monitored as changes catheters that permitted lidocaine solutions to be in the impedance of the fluid column using a 30 instilled into the main lumen of the cuff from a kHz AC bridge circuit, rectified and integrated with port on the connector assembly; results of these ex- a 100-Hz frequency response; most figures here periments are described in a companion paper (23). show at least one such length gauge signal plus an The transformer-coupled amplifier system (1- 10 electrically differentiated velocity trace. The length kHz bandwidth) and spike-triggered averaging calibrations were obtained by measuring the joint techniques are described in detail elsewhere (12). angles occumng at the extreme excursions of a step In the figures shown here, three buffered traces cycle from video stills and trigonometrically deter- (showing averages of signals that occurred before mining the muscle length with a mathematical as well as after the triggering signal) show the av- model of the origin, insertion, and pulley dimen- eraged unit potential shape (as originally recorded sions obtained from all of these muscles in a cadaver with 1- to 10-kHz bandwidth from the DRG mi- specimen of similar size. During walking, the lengths croelectrode), the signal from the proximal tripolar of rectus femoris and both parts of sartorius muscles electrode (FT,center contact positive upward) and are dominated by the hip joint motion, permitting the distal femoral tripolar electrode (FD). From the at least a first-order estimate of length from the ap- time difference between the two negative peaks (de- propriately scaled output of a single implanted termined digitally) and the known separation be- gauge L, (see Fig. 6 and text). However, due to tween the two center contacts, the conduction ve- various differences among animals regarding skel- locity has been calculated. Analysis of the errors to etal structure and internal muscle fiber architecture which this method is subject indicate a 5% uncer- and nonlinearities in the gauges and their curved tainty for the fastest conduction velocities and path motion, these records should be taken as ap- somewhat less for those in the group I1 range (12, proximate and are intended primarily for relative 34). Conduction velocities measured in this manner comparisons. (See Ref. 26 for discussion of differ- may be slightly higher than those determined over ences of muscle lengths determined by implanted longer, more heterogeneous segments of peripheral gauges and calculation from stick figures.) Gains nerve where slowing may occur. The first five units and DC offsets are unchanged for all of the records listed in Table 1 come from animals implanted with shown in any single figure. The sign of the velocity a single tripolar femoral cuff. Conduction velocities 554 LOEB, HOFFER, AND PRATT obtained from the latency between the cuff and the trigger the signal averager (Nicolet 1070 or 1174) DRG spike are less precise, and only approximate to determine femoral nerve projection and con- ranges are given for these units. duction velocity from signals recorded both on-line The EMG signals were obtained from two dif- or, alternatively, from previously taped activity. In- ferent electrode configurations specifically designed stantaneous frequency of these unitary action po- to maximize sampling and minimize cross talk from tentials are shown as frequencygram records (in- the particular muscle geometries occurring in the verse of interspike interval) in the figures here (Bak anterior thigh group. For the vasti and rectus fe- Electronics ISI-I). These appear as dots at the top moris muscles, distributed multipolar spiral elec- of fast vertical deflections that appear as broken trodes were threaded through the central core of lines. Sometimes the filtered microelectrode signal each muscle (13). For the sartorius muscle, there entering the window discriminator is shown un- were usually two separate bipolar patch electrodes derneath, but these wideband records do not repro- designed to sample selectively from the medial (knee duce well on the relatively slow sweep oscillographic flexing) and anterior (knee extending) parts of the traces shown here (Honeywell Visicorder # 1858). broad, flat, superficial muscle (for methods see Ref. On the original records they were very useful in 1 I). Each bipolar patch consisted of a pair of Dacron checking for false acceptances of artifacts or drop- reinforced silicone rubber sheets 0.007" thick (Dow outs resulting from unit amplitude changes. The Corning #501-1) and -- 15 mm square with a 5- unit discrimination was also checked by examining mm-long stranded stainless steel contact in the cen- multiple, overlaid traces of all triggering events to tral region. One square was sutured to the on the window discriminator by using a storage oscil- each side of the sartorius muscle so that the contacts loscope with high gain and fast sweep speed (usually were against the muscle fibers and formed a bipolar 20 pV/division and 0.2 ms/division). sandwich. The dielectric silicone rubber sheet ef- An attempt was made to identify all discriminable fectively screened out current fluxes from under- units recorded by each electrode, according to the lying sources of cross talk such as the rectus femoris following criteria: 1) Presence of a unitary spike- and vastus medialis muscles. EMG signals were re- triggered average response in the femoral nerve corded with 50- to 5,000-Hz bandwidth and are preceding the DRG unit activity. This confirmed displayed here with fullwave rectification and in- the afferent nature of the activity and narrowed tegration into 2-ms wide bins (4). down possible sources to those five muscles supplied The amplified raw microelectrode signals for all by the femoral nerve distal to the nerve cuff plus channels carrying discriminable unit activity were saphenous cutaneous and median articular nerves. taped on an 18-track FM tape recorder (DC- I0 kHz; 2) Conduction velocity based on latency between Sangamo Sabre IV) along with all transducer, EMG, averaged responses in two adjacent tripolar record- and nerve cuff signals plus an indicator of treadmill ing sites on femoral nerve. This was used to distin- speed, stimulus conditions if any, and an IRIG-B guish group I from group I1 afferents (using the 72 time code for synchronization with simultaneous m/s boundary for spindle afferents proposed by videotape records (60 fieldsls, two views from Matthews, Ref. 28). 3) Manipulation of the passive Plumbicon low-persistence cameras with screen limb during anesthesia. This was used to identify splitter). After the recording of all activities and the muscle group of origin of a spindle afferent and manipulations, the animal was anesthetized to per- to distinguish it from other deep receptors including mit probing and manipulation of receptor fields to Golgi tendon organs and joint afferents (see below). ascertain the modality of each discriminable unit, 4) Probing with a hand-held, punctate vibratory using a chronically implanted jugular catheter that probe during anesthesia. This was used to further permitted the rapid induction and close control of localize the receptor bearing structure within a barbiturate anesthesia. In a few cases, the unavail- group of muscles with similar mechanical action ability of a patent jugular catheter forced anesthesia (e.g., the vasti). to be induced via intramuscular ketamine (50-100 The insert diagrams on the unit records show the mg), using amounts necessary to insure passivity of results of the two most useful identification tests, the muscles to the slow stretches used for spindle next to conduction velocity, that were directions of identification. skeletal motion that caused brisk, sustained unit The raw microelectrode signal was further filtered discharge (usually with a phasic initial burst) and and then discriminated by using a time-amplitude the region of highest sensitivity to the vibrating window discriminator and analog delay line (5, 6) probe plus highest frequency of cycle-for-cycle vi- to identify unitary action potentials. Usually, only bration-locked discharge. Joint afferents were the largest waveform was analyzed, although oc- readily identified by their sensitivity to rotational casionally (4 out of 23 units) the next largest spike and other off-axis loading of the knee joint. Cuta- could be cleanly discriminated and turned out to neous afferents (which were the most numerous be a spindle afferent from an instrumented muscle. general class) were identified by their superficial re- The discriminator acceptance pulse was used to ceptive fields. bits suspected of being Golgi tendon SPINDLE ACTIVITY 555 organ afferents were recorded but gave little or no length changes during walking tend to be response to slow stretch of the passive whole muscle, dominated by the larger excursions of the hip as would be expected (37). They were thus indis- joint. At slow to moderate gaits (walking and tinguishable from units that were lost before or slow trotting), RF was recruited with the vasti, during the identification process. A small number of tonically active units with no obvious source of although it tended to reach peak activity later modulation or receptor fields were identified as in in the stance phase than any of the vasti (see previous recordings from similar chronically im- Fig. 5). At faster speeds, it usually had a burst planted electrodes (20). These probably represent of EMG activity in mid to late swing phase, damaged fibers with spontaneous activity; occa- synergistically with the anterior part of sar- sionally femoral nerve averages triggered by such torius. spontaneous spikes indicated an outgoing (anti- Sartorius is a mechanically complex muscle dromic) direction. Electrical stimulation of muscles that originates on the anterior iliac crest and via their chronically implanted EMG electrodes inserts in a continuous sheet extending from provided useful clues regarding the location of the superior margin of the patella, along the spindle receptors. However, both parallel unloading from synergists and early twitch activation rather medial edge of the patella and its ligament, than silencing could be produced with some stimuli and onto the anterior tibia1 ridge. The inner- in some muscles, making this test less than conclu- vating nerve bifurcates early, dividing the sive in this closely mechanically linked system. muscle into an anterior part ( SA-a) with knee extending action and a medial part (SA-m) RESULTS with knee flexing action. During walking, the lengths of both parts tended to be similarly Anatomical and functional organization dominated by the large lever arm acting of the muscles around the hip joint (but spindles in the two The anterior thigh muscles of the cat, al- parts of the muscle were readily differentiated though few in number, are mechanically and by their selective reaction to motion at the kinematically quite diverse (see Fig. 1). The knee joint when the hip was stabilized). The three vastus muscles-intermedius (VI), me- anterior sartorius had two periods of extrafusal dialis (VM), and lateralis (VL)-make up three activity: one during stance and one during of the four heads of the quadriceps, and act as swing, predominately in the later El phase of pure extensors of the knee. They are all the Phillipson step cycle (3 1; Fig. 1B) when its unipinnate (35) and take origin along an ex- combined hip flexion and knee extension ac- tended length of the and insert together tions are both appropriate. The medial sar- on the superior margin bf the patella. The torius tended to have all or most of its activity deepest lying VI is a red muscle composed al- during the early to mid swing phase (depend- most exclusively of type I (presumably slow ing on precise electrode placement), when its twitch) muscle fibers, whereas the more su- combined knee and hip flexion actions are perficially lying parts of VM and VL are pro- both appropriate. We will deal with the an- gressively whiter, with larger percentages of terior and medial parts as if they constituted type I1 (fast twitch) muscle fibers (3). All three separate muscles, although there is no fascia1 muscles were recruited synergistically during plane between them. the stance phase, with a small but consistent phase difference. VI tended to be recruited Spindle aferents identified first, usually just before footfall, and its activity Table I gives the coded name and putative tapered off well before the end of the stance identity of each of the units that we identified phase, whereas VM and particularly VL as spindle afferents from one of the six muscles tended to peak in mid to late stance, and to innervated by the femoral nerve distal to the cease activity more abruptly about 50-100 ms cuff electrode. The identification of a unit sen- before footlift (shown schematically in Fig. sitive only to knee flexion as originating in 1B). The rectus femoris (RF), the fourth head any particular one of the three vasti is based of the quadriceps, is a nonpinnate muscle that largely on the apparent locus of best response takes origin along the wing of the and to weak vibration. This is subjective at best inserts with the vasti on the patella. Its mo- and subject to error; however, given the sim- ment arm about the hip joint, which it flexes, ilarity in mechanical action, fiber architecture, is similar to that about the knee, but its overall and EMG recruitment of the three muscles, 556 LOEB, HOFFER, AND PRATT such an error in- the classification of parent Proprioceptive units with hip and/or knee muscle would not affect the interpretation of motion sensitivity were frequently recorded the spindle activity. A similar consideration without such averages being obtainable even applies to the distinction between SA-a and with 4,000 sweeps; these were all rejected as RF, although the more superficial SA-a muscle potentially arising from noninstrumented provided a better subject for vibratory local- muscles such as tensor . ization of the receptor. A couple of units re- In addition to the figures published here, sponded well to passive hip motion but only Table 1 lists references where we have pub- weakly to knee motion; these presumably lished records of activity of other anterior thigh came from the intermediate region of the SA spindle afferents, usually for conditions other muscle, which attaches to the medial edge of than unperturbed locomotion. the patellar ligament, and hence has no exten- sor or flexoraction on the knee. Vasti muscle spindles From the conduction velocities, it is appar- Figure 3 shows activity from a spindle pn- ent that our method of recording was biased mary afferent typical of those identified as toward the larger, faster fibers, as noted pre- arising from the pure knee extensor vasti viously (20). In addition to the usual biases of group. During the extrafusal muscle activity extracellular recording electrodes, it is possible of the stance phase, the spindle activity was that many group I1 units with conduction ve- modulated between -50 and 200 pps, with a locities slower than 72 m/s were recorded but clear tendency to increase its rate in response not identified as passing through the femoral to a combination of both amplitude and rate nerve cuff. This is because the amplitude of of stretch. At the slower walking speeds, the the extracellular action potential decreases as swing phase activity was much lower despite a power function of decreasing fiber diameter larger and faster stretching of the passive mus- (12, 27), and signal-to-noise increases as the cle during the flexion phase than during stance. square root of the number of averaged sweeps. At the fast trot speed, the even larger, faster flexion movements began to cause more spin- TABLE 1. Spindle afferents recorded dle activity than that seen during the nearly and identified isometric stance phase. Spindle activity was generally quite low or zero during muscle CV, Vib., Gamma shortening in both swing and stance (although ID# Muscle m/s Hz Block Ref. it was occasionally maintained at significant levels through the end-stance extension; see fast walk and end of first stance phase in lower set of traces). The muscle velocities were gen- erally in the range of? 1 rest lengthis, but be- cause of the high angle of pinnation and rel- atively short muscle fibers (-2 cm; Ref. 33, the velocities experienced by muscle fibers and parallel structures such as the muscle spindles J6AlO VM 115 200 X were probably closer to four times that value K12A9 VM 33 NS (if expressed as fiber lengths/s). L3A29 SA-a 121 320 X 23 N5A6 SA-a 119 300 X 23,25 The sequence of events at the bottom of N6A15 SA/RF 103 ND Fig. 3 gives some indication that the pattern N6A27 SA-m 94 210 X 10,25 of discharge observed during walking was not OlOA6 SA-a 120 250 X 25 caused by the transducer properties of a passive S9A12 SA-m 65 NS X spindle lacking in fusimotor drive. At the be- T9B6 VM 109 220 X 23 T12A8 VL 109 330 X 23, 25 ginning of the trace, the animal was sitting T3A15 SA/RF 100 ND with the knee highly flexed, maximally T9A17 SA-m 98 220 X 23 stretching the vasti muscles but causing only T9A22 VM 111 400 X 23 low-frequency, irregular afferent activity. As T2B8 V? 45 ND the animal built up muscle tone in preparation NS, not sensitive; ND, not determined; CV, conduction for extending the knee and standing, the ac- velocity; Vib., vibration. tivity of the afferent rose dramatically but SPINDLE ACTIVITY

Slow Walk-2Ocmls Fast Walk-85crnls f rot-140crnIr

---- medial

VM a ' L VL ...... SA-a A&& . +&- dC.. ,:a&L -- Stance - -

Stand Up and Walk 45cmls

Spike Triggered Avereoe . :..! I .,... nz1024

FTG. 3. Records from vastus medialis spindle primary afferent T9A22. The sketch (upper leJ) shows the region of maximal sensitivity to vibration (shaded region) that the unit followed at up to 400 Hz. The unit was driven by flexion of the knee joint only. The spike-triggered averaged records at lower le$ from 1,024 sweeps revealed femoral nerve action potentials whose interval corresponded to a conduction velocity of I I I m/s. For the activities shown, traces from top down give the unit frequencygram (instantaneous rate of firing), the velocity of the vasti muscles (lengthening is positive), the length of the vasti muscles (stretched is positive), the rectified EMGs from vastus medialis (VM), vastus lateralis (VL), and anterior sartorius (SA-a), the stance periods when the ipsilateral hindfoot was in contact with the treadmill surface, the force at the patellar ligament (F,), and the treadmill speed indicator (5 cm forward rnotion/tick mark). See Ref. 26 for further discussion of length gauge properties.

continued irregularly. During the slow, steady amounts of activity during relatively rapid active muscle shortening of standing up, the passive stretch in the flexion (F) phase of spindle maintained an almost constant dis- swing. The spindles generated little or no ac- charge at the level that eventually was present tivity during passive muscle shortening in the during the stance phase of the walking. extension (El) phase of swing or during rapid, In general, spindle activity during loco- mostly passive muscle shortening near the end motion from all of the vasti muscles was quite of stance phase. similar. It consisted of vigorous activity during most of the stance phase when the muscle Sartorius muscle spindles generated active tension under slowly length- The activity patterns of muscle spindle pri- ening or near isometric conditions, with lesser maries from sartorius muscle were consider- LOEB, HOFFER, AND PRATT

medial

Spike Triggerea Average n-1024

DRC --.J) &-

Fp A-.~Stance - FD .

FIG. 4. Records from anterior sartorius spindle primary 010A6, sensitive to knee flexion and hip extension as shown in the insert sketch, with 120 m/s conduction velocity from the spike-triggered average at lower left. (Slight phase lag of electronically derived velocity trace is artifactual.) Raw microelectrode recording is shown under the unit frequen- cygram. ably more complex and diverse, even when large excursions. Second, the animal was only the division into anatomically distinct anterior 6 days postoperative at the time of this re- and medial parts was taken into account. Fig- cording and began to limp progressively more ure 4 shows the most common pattern, with severely in the later records Band C. The pos- activity during both the swing and stance tural shift can be seen in the decreased am- phases, despite the fact that the muscle was plitude and velocity of the hip joint excursions rapidly shortening during swing and rapidly and a subtle decrease in the rate of rise of the lengthening during stance. The only period of patellar force. The changes in the spindle ac- silence was at the end of stance and early tivity were much less than those in the muscle swing, when the niuscle had reached maximal excursions, with about a 50% decrease during length and began passively shortening prior to the stance phase lengthening and no consistent the active shortening in E, . change during the reduced swing phase short- Figure 5 shows another spindle primary ening. from the same anterior part of the muscle, Figure 6 shows a spindle secondary ending again with generally similar levels of activity (conduction velocity 65 m/s) located in the during both the lengthening stance phase and medial part of the sartorius muscle. The peak the shortening swing phase. This unit is of fur- firing rate rarely exceeded 100 pps and was ther interest for two reasons. First, the stance- relatively smoothly modulated (perhaps even phase activity had a sudden drop during the more so than shown in the frequencygram that fastest rates of lengthening (see arrow in record appears to have been influenced by a few A), a feature that we have noted previously missed action potentials). The length gauge (18) and that was not uncommon during such across the rectus femoris and anterior sartorius SPINDLE ACTIVITY

N5A6 SA-a Ia medial Walking with Limp A B

Spike Triggered Averege n-1024

DUG -qr-

.: !?. a*.. .. VI ; .. & Iw+r~;C.. FD ---\I"----- I ,.oms I Stance - - - * T ...... 50cm18 35cmls 4Ocmla 11 10m18 I 1.08 I

FIG. 5. Records from anterior sartorius spindle primary N5A6, sensitive to knee flexion and hip extension, with 119 m/s conduction velocity. Relatively normal walking in A reverted to a limp in B and C with progressively less motion at the hip and knee as the animal (6 days postoperative) became tired. Arrow in A indicates pause possibly related to yielding of intrafusal structures. muscles (LR)would not accurately reflect the record from the vastus gauge (Lv) was scaled effect of the knee joint motion for this medial appropriately and subtracted from the LR out- part of the sartorius, which flexes rather than put. The resultant LR-LVis shown with a cal- extends the knee. A crude attempt to estimate ibration based on trigonometric reconstruc- the maximal effect of this factor is shown in tion oflhe excursions of this part of the muscle the uppermost length trace, in which the length from video stick figures. When interpreted us-

S9A12 SA-m 11 Slow Walk-40cmls Fast Walk-80cmls

Lv v SA-m -

HG. 6. Records from medial sartorius spindle secondary S9A12, sensitive to knee extension and hip flexion, with 65 m/s conduction velocity. See text for explanation of length traces. LOEB, HOFFER, AND PRATT

E8A8

N6A27

E9A6

0 10A6

L3A29

N5A6

T9A17

length

force stance

HG. 7. Survey of records from spindle primary afferents in various animals walking at a similar speed, correlated with their apparent receptor location in the sartorius muscle complex (based on vibratory probe responses). See Table I for location based on response to manipulation of knee, which should be more reliable at the narrow end of the muscle near its origin. The frequencygram sketches are an approximation of the typical patterns seen in longer sequences of normal walking. Length record from LR gauge is most representative of anterior sartorius motion; knee motion during walking is phased in a way that tends to increase the range and peak velocities for the medial sartorius. ing this net length, the velocity dependence of other afferents reached their peak activity dur- the secondary afferent was somewhat lower ing the peak velocity of shortening. In general, than that of the primary endings. Note that firing rates rarely exceeded 200 pps and, unless the secondary did not fall silent when the off entirely, were rarely less than 20-50 pps. length peaked at footlift (zero velocity) but only when the ensuing shortening reached a DISCUSSION fairly high negative velocity. The mid-swing- Two factors mitigate against attempts to phase activity clearly occurred when the mus- explain these and other published records of cle was still shortening as it approached its spindle afferent activity by invoking particular minimal length and could not be attributed to fusimotor effects. First, spindles can vary sig- El phase stretching, which was not accom- nificantly in the deefferented sensitivity to panied by any significant discharge of the end- particular velocities and extents of applied ing. The higher velocities but generally similar length changes and in the mechanical linkages excursions at the faster gait had no significant that convey such length changes from the ten- effect on either the timing or the amplitude of dons of their muscles to the spindles them- this spindle secondary's activity. selves. Neither of these factors has been ex- Figure 7 is a sampler of spindle primary ac- amined for the spindles of the particular mus- tivity during similar step cycles of walking at cles that were the subject of this study. Second, moderate speeds, together with the apparent the range of quantitative and qualitative effects location of each receptor structure based on of fusimotor activity on the transducer prop- manipulation and vibration. There seem to be erties of spindle afferents is just beginning to no clear anatomic patterns to account for the be understood (e.g., 7, 14, 15). Attempts to various temporal patterns. Some afferents study the interactions of such effects with each generated more activity during stance-phase other and with the complex dynamics of nor- lengthening than during swing-phase short- mal limb traiectories and extrafusal muscle ening, and some had the reverse. Some of the activity face iormidable technical and theo- swing-phase activity occurred relatively late, retical obstacles ( 17). when the muscle was quite short, but the ve- However, it is possible to infer from afferent locity of shortening was decreasing, whereas activity patterns alone that there was at least SPINDLE some fusimotor influence acting on almost all vation (9), it must be assumed that the dy- anterior thigh spindles studied. For the vasti namic and static fusimotor systems, with their muscles, this is indicated by the larger response very different effects, can be independently of the afferents to small stretches during stance turned on or off. That is, some phases of alpha (active lengthening) than to much larger motoneuron recruitment would have to be stretches during swing (passive lengthening). accompanied by coactivation of gamma static For the biarticular muscles, fusimotor action motoneurons, whereas other phases would is indicated by afferent activity during rapid have to be accompanied by gamma dynamic muscle shortening, which is never seen in deef- motoneurons, probably to the exclusion of ferented spindle afferents. In the accompa- gamma statics. This possibility becomes more nying paper (23) we present data from func- consistent with traditional ideas about fixed tional deefferentation experiments using li- recruitment order in motoneuron pools in docaine blockade to provide some qualitative light of the finding that the two EMG bursts and quantitative estimates of these fusimotor occurring in each step cycle in anterior sar- effects. torius are the work of two separate groups of How far can this apparent fusimotor influ- independently recruited alpha motoneurons ence be dissected before resorting to pure con- (1 I). However, the two companion papers jecture? The responses of the vastus medialis shed further light on the amount and com- spindle primary during walking in Fig. 3 were plexity of fusimotor influence on spindle af- highly velocity dependent, suggesting dynamic ferent activity and reveal circumstances in fusimotor influence, but the initial sequence which the extrafusal activity appears to be in- as the animal stood up before walking would dependently modulated. For all of these rea- be more consistent with biasing by static fu- sons, it seems unwise to presume that the re- simotor activity (note steady, somewhat irreg- cruitment of various parts of the fusimotor ular activity during shortening). On different apparatus is determined simply and invariably grounds, it seems clear that the activity of sar- by activation levels in the alpha motoneuron torius spindle afferents during swing phase pool. shortening (e.g., Figs. 4 and 5) must be the We propose that during voluntary move- result of static fusimotor activity, but it is not ments, the fusimotor system functions to con- clear from these records whether the stance tinuously adjust the sensitivity and bias of the phase activity during stretching is consistent spindle afferents to improve the transduction with passive spindle properties, static fusi- of the particular range of mechanical events motor, or dynamic fusimotor effects. The anticipated by the animal. The evolving use pause in the middle of a large stretch noted in of limbs for multiple, independent, finely Fig. 5A suggests a sudden yielding of a stiff controlled tasks in mammals (as opposed to intrafusal element with a slow recovery time, lower vertebrates) has been accompanied by perhaps a bag, fiber under dynamic fusimotor the separation of intra- and extrafusal control activation, but this is again conjectural. into at least three types of motoneurons having If one considers all of the possible interactive very different effects: alpha motoneurons effects of temporally modulated fusimotor ac- modulate force output without influencing tivity, many combinations will be consistent spindle sensitivity, gamma static motoneurons with these data. Perhaps more importantly, maintain spindle bias during muscle short- many combinations are not likely. The sudden ening (extending dynamic range) without silencing of most spindles following length adding to force output, and gamma dynamic peaks in late stance indicates little or no static motoneurons enhance spindle sensitivity to fusimotor activity, whereas sustained discharge small length and velocity perturbations (nar- during rapid shortening requires such static rowing dynamic range). Various combinations activity. It is clear that the similar range of of these systems (including hardwired com- firing rate noted in all of these spindle pri- binations offered by beta motoneurons) can maries during various periods of active parent be and apparently are used to cope with such muscle use cannot have been achieved by the varied kinematic behavior as we have noted same fusimotor program in all cases. In order in the various anterior thigh muscles during to interpret all of these patterns of spindle ac- various phases of locomotion. tivity as resulting from alpha-gamma coacti- In this discussion, it seems more useful to 562 LOEB, HOFFER, AND PRAn take a broader view of the phenomenology of tion of many afferents by postsynaptic neurons spindle afferent traffic and to consider its pos- solves this problem only at the expense of low sible interpretation by the spinal cord and the signal-to-noise levels or degraded spatial res- higher motor control structures that presum- olution (25a). If the central nervous system is ably rely on these signals to guide and refine interested in optimizing the sensitivity of its their output decisions. In that context, the motor control apparatus to unexpected per- most striking general finding is that almost all turbations in the limb trajectory (either short- of the spindle afferents from which we re- ening or lengthening the muscle), it must do corded had similarly well-modulated activity two things: 1)It must keep the spindle afferent during the periods of extrafusal activity of their firing rates within this window of optimal firing parent muscles. Spike rates rarely exceeded rates. 2) It must adjust afferent sensitivity so -200 pps and rarely fell below 20-50 pps that the range of perturbations anticipated (somewhat lower for secondaries), although during the movement causes maximal mod- sustained low-frequency activity was a typical ulation of afferent discharge within this feature of these same muscle spindles when window. the animal was lying quietly or lightly anes- The different independent effects of the thetized with the limb at an intermediate an- static and dynamic fusimotor neurons on gle. A similar tendency has been noted for spindle afferent dynamic range noted above spindles in a wide variety of muscles and mo- are particularly well suited to these tasks, but tor tasks (for review see Ref. 19). The sequence only if the central nervous system has inde- shown at the bottom of Fig. 3 was typical of pendent control of the various types of mo- the systematic biasing of activity that accom- toneurons. panied the onset of locomotor activity in the Such an interpretation of fusimotor activity anterior thigh muscles. In a perhaps related suggests a specific division of labor between feature, the activity of these afferents changed the primary and secondary endings. The same very little with changes in the speed or extent programming of the primary endings that of the joint angle excursions and/or changes might optimize instantaneous sensitivity to in parent muscle EMG during locomotion unexpected perturbations of limb trajectory made by voluntary actions of the animal. This can also be expected to greatly complicate the suggests that such voluntary changes in the extraction of kinesthetic information about extrafusal motor program were accompanied posture. Information about wide-ranging but by compensatory changes in the intrafusal slowly changing aspects of limb and trunk po- program, consistent with the servo-control sition is probably an important output of the notion of using sensory feedback to provide muscle spindles, in concert with joint and skin an error signal to the motor controller (see stretch receptors. The secondary endings, with Ref. 36). their less rapidly modulated firing rates and All of this is consistent with a systematic more linear responses to length, might provide optimization of instantaneous transducer sen- a better input signal to a central integrating sitivity to make the best use of the informa- circuit with relatively long time constant for tion-conveying properties of an afferent data keeping track of overall limb position. Of stream consisting of all-or-none action poten- course, at least part of the fusimotor control tials propagating along an axon (for quanti- system that influences the primary endings is tative analysis, see Ref. 25a). Firing rates above shared with the secondaries (the chain fibers 200 pps appear to be difficult to sustain; the innervated primarily by gamma static moto- rare instances of such rates in our own data neurons). However, the transducer sensitivity (e.g., Figs. 4 and 5) and similar records in other effects of these fibers tend to improve rather muscles (22, 32, 33) appear intermittent and than degrade linearity of the response to length noisy, tending to degrade the information that changes by reducing velocity-dependent effects can be extracted from the afferent signal. Fir- (16) and extending the useful range of the ing rates below 50 pps cannot signal instan- transducer into the negative velocity (short- taneous changes in mechanical input simply ening) regions (2). because they must be integrated over at least The spindle afferent activity studied here their interpulse intervals (20 ms or longer) be- includes examples of muscles classified as ex- fore changes can be appreciated. The integra- tensors (vasti and rectus fernoris) and as flexors SPINDLE ACTIVITY 563

(medial sartorius). A difference in the spindle For similar reasons, the bifunctional muscle activity between ankle extensors and flexors studied here, sartorius anterior, is of interest has been noted previously (22) in that the ex- because it has periods of both active shortening tensors (triceps surae) had relatively simple typical of a flexor and active lengthing typical patterns of activity related to stretch whereas of an extensor (although somewhat exagger- flexor muscle spindles (tibialis anterior and ated). As was noted above, the two electro- ) generated activity during myographic periods of activity in the anterior both active muscle shortening and passive part of the sartorius muscle appear to be gen- stretch. However, the anatomical or physio- erated by two independent motoneuron logical classification of a muscle as a "flexor" groups (1 1). The complex and rather individ- or an "extensor" per se may not be the deter- ualized activity patterns of the sartorius muscle mining factor in the control or use of that spindles may reflect the need to provide ap- muscle's proprioceptors. The kinematics of propriate proprioceptive feedback for these muscle use revealed in this study are consistent two separate motoneuron groups. It may also with the distinction noted in those and other reflect the complex interaction of intrafusal studies of the ankle muscles (1, 38) that the effects of gamma and beta motoneurons (8) antigravity extensors usually were active dur- recruited in concert with these functionally ing lengthening or isometric conditions, separate alpha motoneuron groups. Resolu- whereas most flexors were active while rapidly tion of these questions will require a detailed shortening. This functional dichotomy may be knowledge of the neural projection patterns more important to the fusimotor control of among individual alpha, beta, and gamma the spindle than the more conventional clas- motoneurons and the spindle afferents pro- sification of flexors versus extensors. In this ducing those various patterns during loco- context, studies of muscles with "crossed motion. roles" such as extensors that actively shorten (e.g., monarticular ) or flexors that Current address of Dr. Hoffer: Department of Clinical actively lengthen (e.g., tibialis anterior in Neurosciences, Uliversity of Calgary, Faculty of Medicine, plantigrade walkers) should help to establish Calgary, Alberta, T2N IN4, Canada. which are the important factors to the or- Received 22 June 1984; accepted in final form 25 March ganism. 1985.

REFERENCES

1. ABRAHAM,L. D. AND LOEB,G. E. The distal hindlimb 8. BOYD,I. A., GLADDEN,M. H., MCWILLIAM,P. N., musculature of the cat: patterns of normal use. Exp. AND WARD,J. Control of dynamic and statis nuclear Brain Res. 58: 580-593, 1985. bag fibres and nuclear chain fibres by gamma and 2. APPENTENG,K., PROCHAZKA,A., PROSKE, U., AND beta axons in isolated cat muscle spindles. J. Physiol. WAND,P. Effect of fusimotor stimulation on Ia dis- London 265: 133-162, 1977. charge during shortening of cat at dif- 9. GRANIT,R. The functional role of the muscle spin- ferent speeds. J. Physiol. London 329: 509-526, 1982. dles-facts and hypothesis. Brain 98: 53 1-556, 1975. 3. ARIANO,M. A., ARMSTRONG,R. B., AND EDGERTON, 10. HOFFER, J. A. AND LOEB,G. E. A technique for re- V. R. Hindlimb muscle fiber populations of five versible fusimotor blockade during chronic recording mammals. J. Histochem. Cytochem. 21: 51-55, 1973. from spindle afferents in walking cats. Exp. Brain Res. 4. BAK,M. J. AND LOEB,G. E. A pulsed integrator for Suppl. 7: 272-279, 1983. EMG analysis. Electroencephalogr. Clin. Neurophys- 11. HOFFER,J. A., LOEB,G. E., O'DONOVAN,M. J., AND iol. 47: 738-741, 1979. PRATT,C. A. Unitary activity patterns during walking 5. BAK, M. J. AND SCHMIDT, E. M. An analog delay confirm the existence of two functionally distinct circuit for on-line visual confirmation of discriminated classes of sartorius motoneurones in cats. J. Physiol. neuroelectnc signals. IEEE Trans. Biomed. Eng. 24: London 308: 20P, 1980. 69-7 1. 1977. 12. HOFFER, J. A., LOEB,G. E., AND PRATT,C. A. Ulitaq 6. BAK, M. J. AND SCHMIDT,E. M. An improved time- axonal conduction velocity from averaged nerve cuff amplitude window discriminator. IEEE Trans. electrodes in cats. J. Neurosci. Methods 4: 21 1-225, Biomed. Eng. 24: 486-489, 1977. 1981. 7. BOYD,T. A. The action of the three types of intrafusal 13. HOFFER, J. A., O'DONOVAN, M. J., PRAlT, C. A., fibre in isolated cat muscle spindles on the dynamic AND LOEB,G. E. Discharge patterns in hindlimb mo- and length sensitivities of primary and secondary sen- toneurons during normal cat locomotion. Science 2 13: sory endings. In: Muscle Receptors and Movement. 466-468, 1981. edited by A. Taylor and A. Prochazka. London: Mac- 14. HULLIGER,J. AND NOTH,J. Static and dynamic fu- millan, 198 1, p. 17-32. simotor interaction and the possibility of multiple 564 LOEB, HOFFER, AND PRA'IT

pace-makers operating in the cat muscIe spindle. Brain The Muscle Spindle. London: Macmillan. In press. Res. 173: 21-28, 1979. 26. bEB, G. E., WALMSLEY,B., AND DUYSENS,J. Ob- 15. HULLIGER,M., MATTHEWS,P. B. C., AND NOTH, J. taining proprioceptive information from natural Effects of combining static and dynamic fusimotor limbs: implantable transducers vs. somatosensory stimulation on the response of the muscle spindle pri- neuron recordings. In: Physical Sensors for Biomedical mary ending to sinusoidal stretching. J. Physiol. Lon- Applications. Proc. of Workshop on Solid State Phys- don 267: 839-856, 1977. ical Sensors for Biomedical Application, edited by 16. HULLIGER,J. AND NOTH,J. Static and dynamic fu- M. R. Neuman. Boca Raton, FL: CRC, 1980, p. 135- simotor interaction and the possibility of multiple pace 149. makers operating in the cat muscle spindle. Brain Res. 27. MARKS,W. B. AND LOEB,G. E. Action currents, in- 173: 21-28, 1979. ternodal potentials, and extracellular records of my- 17. HULLIGER,M. AND FROCHAZKA,A. A new simula- elinated mammalian nerve fibers derived from node tion method to deduce fusimotor activity from afferent potentials. Biophys. J. 16: 655-668, 1976. discharge recorded in freely moving cats. J. Neurosci. 28. MATTHEWS,P. B. C. Mammalian Muscle Receptors Methods 8: 197-204, 1983. and Their Central Actions. London: Arnold, 1972. 18. LOEB, G. E. Somatosensory unit input to the spinal 29. MURTHY,K. S. K. Vertebrate fusimotor neurones cord during normal walking. Can. J. Physiol. Phar- and their influences on motor behaviour. Prog. Neu- macol. 59: 627-635, 1981. robiol. 1 I: 249-307, 1978. 19. LOEB,G. E. The control and response of muscle spin- 30. O'DONOVAN,M. J., HOFFER,J. A., AND bEB, G.E. dles during normally executed motor tasks. Exercise Physiological characterization of motor unit properties Sport Sci. Rev. 12: 157-204, 1984. in intact cats. J. Neurosci. Methods 7: 137- 149, 1983. 20. LOEB,G. E., BAK,M. J., AND DWSENS, J. Long-term unit recording from somatosensory neurons in the 3 1. PHILI~PSON,M. L'Autonomie et la centralisation dans spinal ganglia of the freely walking cat. Science 197: le systeme nerveux des animaux. Trav. Lab. Physiol. 1192-1194, 1977. Iut. 7: 1-208, 1905. 21. LOEB, G. E., BAK, M. J., SALCMAN,M., AND 32. FROCHAZKA,A., STEPHENS,J. A., AND WAND, P. SCHMIDT,E. M. Parylene as a chronically stable, re- Muscle spindle discharge in normal and obstructed producible microelectrode insulator. IEEE Trans. movements. J. Physiol. London 287: 57-66, 1979. Biomed. Eng. 24: 121-128, 1977. 33. PROCHAZKA,A., WESTERMAN,R. A., AND ZICCONE, 22. LOEB, G. E. AND DUYSENS, J. Activity patterns in S. P. la afferent activity during a variety of voluntary individual hindlimb primary and secondary muscle movements in the cat. J. Physiol. London 268: 423- spindle afferents during normal movements in unre- 448, 1977. strained cats. J. Neurophysiol. 42: 4204, 1979. 34. RINDOS, A. J., ~EB,G. E., AND LEVITAN,H. Con- 23. LOEB,G. E. AND HOFFER,J. A. Activity of spindle duction velocity changes along lumbar primary affer- aKerents from cat anterior thigh muscles. 11. Effects ents in cats. Exp. Neurol. 86: 208-226, 1984. of fusimotor blockade. J. Neurophysiol. 54: 565-577, 35. SACKS, R. AND ROY, R. R. Architecture of the hind 1985. limb muscles of the cat: functional significance. J. 24. LOEB,G. E. AND HOFFER,J. A. Muscle spindle func- Morphol. 173: 185-195, 1982. tion during normal and perturbed locomotion in cats. 36. STEIN, R. B. The peripheral control of movement. In: Muscle Receptors and Movement, edited by A. Physiol. Rev. 54: 215-243, 1974. Taylor and A. Prochazka. London: Macmillan, 198 1, 37. STEPHENS,J. A,, REINKING,R. M., AND STUART, p. 219-228. D. G. The tendon organs of cat medial gastrocnemius: 25. ~EB,G. E., HOFFER,J. A., AND MARKS,W. B. Ac- responses to active and passive forces as a function pf tivity of spindle afferents from cat anterior thigh mus- muscle length. J. Neurophysiol. 38: 1217-1231, 1975. cles. 111. Effects of external stimuli. J. Neurophysiol. 38. WALMSLEY,3.. HODGSON,J. A., AND BURKE,R. E. 54: 578-591, 1985. Form produced by medial gastrocnemius and soleus 25a. LOEB,G. E. AND MARKS,W. B. Optimal control muscles during locomotion in freely moving cats. J. principles for sensory transducers. In: Proc. Inl. Symp.: Neurophysiol. 41: 1203-1216, 1978.