Scratch responses in normal cats: hindlimb kinematics and muscle synergies.

JOURNALOFNEUROPHYSIOLOGY Vol. 64, No. 6, December

1990.

Printed

in U.S.A.

Scratch Responses in Normal Cats: Hindlimb Kinematics and Muscle Synergies PATRICIA CARLSON KUHTA AND JUDITH L. SMITH Laboratory of Neuromotor Control, Department of Kinesiology and the Brain Research Institute, University of California, Los Angeles, California 90024-l 568 SUMMARY

AND

CONCLUSIONS

1. Scratch responses evoked by a tactile stimulus applied to the outer ear canal were characterized in nine adult cats. Chronic electromyographic (EMG) electrodes were surgically implanted in selected flexor and extensor muscles of the hip, knee, and ankle joints to determine patterns of muscle activity during scratching. In some trials EMG records were synchronized with kinematic data obtained by digitizing high-speed tine film, and in one cat, medial gastrocnemius (MG) tendon forces were recorded along with EMG. For analysis the response was divided into three components: the approach, cyclic, and return periods. Usually scratch responses were initiated with the cat in a sitting position, but in some trials the animal initiated the response from a standing or lying posture. 2. During the approach period the hindlimb ipsilateral to the stimulated ear was lifted diagonally toward the head by a combination of hip and ankle flexion with knee extension. Hindlimb motions during the approach period were associated with sustained EMG activity in hip-flexor, knee-extensor (occasionally), and ankle-flexor muscles. Initial hindlimb motions were typically preceded by head movements toward the hindpaw, and at the end of the approach period, the head was tilted downward with the stimulated pinna lower than the contralateral ear. During the return period movements were basically the reverse of the approach period, with the hindpaw returning to the ground and the head moving away from the hindlimb. 3. During the cyclic period the number of cycles per response varied widely from 1 to 60 cycles with an average of 13 cycles, and cycle frequency ranged from 4 to 8 cycles/s, with a mean of 5.6 cycles/s. During each cycle the paw trajectory followed a fairly circular path, and the cycle was defined by three phases: precontact, contact, and postcontact. On average the contact phase occupied -50% of the cycle and was characterized by extensor muscle activity and extension at the hip, knee, and ankle joints. The hindpaw contacted the pinna or neck at the base of the pinna throughout the contact phase, and paw contact typically resulted in a rostra1 motion of the head as the hindlimb extended. 4. The postcontact phase constituted -24% of scratch cycle and was usually initiated by the onset of knee flexion. Ankle and then hip flexion followed knee flexion, and flexor muscles were active during the postcontact phase as the paw was withdrawn from the head. The precontact phase constituted -26% of scratch cycle and was initiated by knee joint extension and knee-extensor activity. Ankle-extensor activity and ankle extension followed closely, and the paw moved towards the stimulated ear. During the precontact phase the hip continued to flex, and hip-extensor activity was initiated at or immediately after paw contact. 5. Muscle synergies for scratching were generally characterized by alternate flexor [tibialis anterior (TA), extensor digitorum longus (EDL), and iliopsoas (IP)] and extensor [lateral gastrocnemius (LG), soleus (SOL), MG, vastus lateralis (VL), anterior biceps femoris (ABF), and gluteus medius (GM)] activity; how0022-3077/90

$1 SO Copyright

ever, within each synergy, recruitment of different muscles was not synchronous. Onset of extensor activity, for example, was sequential: ankle-, knee-, and hip-extensor activity; this recruitment matched the order of joint extension during the precontact phase. On average, extensor muscles were active for -40% of the cycle, whereas flexor activity occupied -50% of the cycle, and increases in TA and LG burst durations were related to cycle-period increases. Activity of the semitendinosus (ST), a bifunctional muscle, spanned the extensor-flexor transition and was not classified as part of either synergy. 6. Brief extensor bursts and low extensor-tendon forces occurred when paw contact was light; conversely, extensor bursts were prolonged and tendon forces increased when firm contact occurred. These data suggest that feedback during the initial phase of contact may prolong extensor activity and delay the onset of flexor activity. Similarly, comparisons of our records with those from no-contact, air-scratch cycles-typical of decerebrate cats studied by others- reveal marked differences in the relative timing of flexor and extensor activity. In normal scratching flexor and extensor muscles have bursts of similar duration, but in air scratching flexor bursts are -8-10 times longer than extensor bursts. Taken together, these data suggest that the relative timing of extensor-flexor activity is modulated by motionand posture-related feedback from limb and head receptors, or by input from descending pathways controlled by supraspinal centers.

INTRODUCTION

The pioneering work of Sherrington ( 1906, 19 10, 1917) and Graham Brown (1909) on the scratch reflex of various mammals (cat, dog, and guinea pig) focused attention on cyclic and stereotypic hindlimb motions other than locomotion. Although the behavior was initially classified as a reflex, both Graham Brown and Sherrington recognized elements of central control, and Sherrington (1906) observed that scratching was elicited even after the hindlimb of the spinal-transected dog had been deafferented. Nearly 50 yr later, Jankowska (1959) reported that the cat scratch response could also be elicited after hindlimb deafferentation (also see Goldberger and Murray 1974). Jankowska observed, however, that paw motion occurred “only in the air without touching the skin.” Nonetheless, from an analysis of tine film records, she noted that movements of the deafferented hindlimb “were rhythmic and their amplitude and frequency were not very different” from those of normal scratches. In the 1970s a group of Moscow-based scientists published a series of studies on scratching in the decerebrate

0 1990 The American Physiological Society

Downloaded from www.physiology.org/journal/jn at Tulane University (129.081.226.078) on February 12, 2019.

1653

1654

KUHTA

AND

(also decapitate) cat. In the initial study Deliagina et al. (1975) showed that hindlimb scratching could be elicited by a combination of chemical and electrical stimulation of upper cervical segments (also see Domer and Feldberg 1960; Feldberg and Fleischhauer 1960). Seconds after stimulation began, the “postural component” of the response was initiated, and the hindlimb was elevated by flexor activity. Subsequent cyclic (3-4 cycles/s) scratching was characterized by alternating flexor and extensor activity. During each scratch cycle flexor activity occupied - 80% of the cycle and extensor activity only 20%; thus a long (-250 to 300 ms) bout of flexor activity held the limb in flexion, whereas a short (-40 to 50 ms) extensor burst produced a rapid downward excursion. Neither the electromyographic (EMG) pattern nor the limb kinematics were altered by extensive hindlimb deafferentation. Deliagina and her colleagues ( 1975) also showed that a fictive motor program, recorded from muscle nerves with the decerebrate cat paralyzed by gallamine triethiodide (Flaxedil), was similar to EMG records of actual scratching. The fictive-decerebrate preparation was then used to assess scratch-induced recruitment of hindlimb motoneurons (Berkinblit et al. 1980; Deliagina et al. 198 l), lumbosacral interneurons (Berkinblit et al. 1978a,b; Deliagina et al. 1983), Ia-inhibitory interneurons (Deliagina and Orlovsky 1980), Renshaw cells (Deliagina and Feldman 198 l), muscle spindle afferents (Feldman et al. 1977), and various areas of the cerebellum (see Arshavsky et al. 1986 review). These comprehensive studies of fictive scratching by the Moscow group (see review by Gelfand et al. 1988) and those of a Kiev-based group (Baev 198 1; Baev et al. 198 1; Shimanskii 1988; Shimanskii and Baev 1987) led to the conclusion that there is a central generator for scratching within lumbosacral segments of the spinal cord. Lacking from the substantial literature on the cat scratch response is an assessment of scratching in normal cats; thus no comparisons of the motor output between normal and decerebrate cats are available. For locomotion such comparisons have been essential in determining aspects of the motor program not specified by spinal generators (Grillner 198 1). We suspected that the motor patterns of decerebrate and normal scratching might be different, because the decerebrate cat’s position during scratching is rigidly constrained by head and trunk fixation; moreover, cutaneous input is absent and motion-related feedback is reduced because the paw cannot contact the stimulated site. Decerebrate scratching is not “actual” scratching as it is often defined; it is air scratching as the paw cycles in the air at midtrunk. How similar, then, are air-scratch cycles to those in which postural orientations of the head and trunk permit the paw to contact a stimulated site on the pinna? A few studies of scratching behaviors in normal cats are published. Sherrington (19 17), for example, provided an anecdotal account of scratching in adult cats, and Jankowska ( 1959) assessednormal and conditional scratching in normal cats and included a few samples of hindlimb trajectories for a few scratch cycles. Recently, scratching was included as one of three cyclic actions (with walking and paw shaking) to assesssynergistic actions of selected ankle and toe muscles (Abraham and Loeb 1985; O’Donovan et al. 1982). Data from these studies, however, have

SMITH

not provided a quantitative assessment of EMG or hindlimb kinematics for the normal scratch cycle. In a recent study from our laboratory, Bradley and Smith ( 1988) assessedthe development of scratching in kittens by recording EMG from three hindlimb muscles. Unlike data from air- and fictive-scratch cycles of the decerebrate cat, the kitten’s scratch cycle was comprised of alternating flexor and extensor bursts of nearly equal duration. These EMG data provided the first indication that some aspects of the scratch response in normal and decerebrate cats were different. Here, we extend the Bradley and Smith study by providing a more extensive EMG analysis and by including a kinematic analysis of the scratch response for normal, adult cats. We found that a postural set, particularly the positioning of the head before the hindlimb was elevated to scratch, was an important behavioral prerequisite. On average, the rate of scratching was slightly faster in normal than decerebrate cats, and the normal cat’s paw contacted the stimulated site during each cycle. Although flexor and extensor synergies at the hip, knee, and ankle joints were similar to those described for air- and fictive-scratch cycles, the relative timing of flexor and extensor bursts was not. Scratch cycles of normal cats were characterized by flexor and extensor bursts of similar duration; this is in marked contrast to air- or fictive-scratch cycles of the decerebrate cat, which are characterized by long-flexor and brief-extensor bursts. Differences in the relative timing of flexor and extensor activity may result from feedback modulation during paw contact, as well as input from supraspinal centers related to head and neck postures. Our findings are summarized in two abstracts (Carlson Kuhta and Smith 1988; Carlson Kuhta et al. 1989). METHODS

Subjects and surgical preparations Nine laboratory-raised, adult female cats (2-3 kg) were subjects. Fine-wire electrodes were surgically implanted in selected hindlimb muscles under aseptic conditions and pentobarbital sodium anesthesia (30 mg/kg iv). Before anesthetic induction three preanesthetics were administered: atropine sulfate (0.05 mg/kg SC),acepromazine maelate (0.2 mg/kg SC),and ketamine hydrochloride (13 mg/kg im). Immediate postoperative care included administration of a respiratory stimulant [Doxapran hydrochloride (Dopram); 5 mg/kg im], fluids, an analgesic [Pentazocine lactate (Talwin); 1-3 mg/kg im], and an oral antibiotic were given as necessary to facilitate postsurgical recovery. Details of implantation procedures are found in Betts et al. (1976). Briefly, multistranded, bipolar wires were led subcutaneously from a skullmounted, multipin connector to the surgically exposed muscle. A I- to 2 mm section of Teflon insulation was removed from each wire electrode; the exposed area was centered within the muscle, and the distal end of each wire was sutured to fascia overlying the muscles or directly to superficial fibers. Implanted muscles included an ankle flexor, tibialis anterior (TA, 7 cats); an ankle flexor and toe extensor, extensor digitorum longus (EDL, 1 cat; see acknowledgements); two ankle extensors, soleus (SOL, 3 cats) and lateral gastrocnemius (LG, 5 cats); a knee extensor, vastus lateralis (VL, 6 cats); a knee flexor and hip extensor, semitendinosus (ST, 2 cats); a hip flexor, iliopsoas (IP, 4 cats), and two hip extensors, gluteus medius (GM, 3 cats) and the anterior compartment (defined by English and Weeks 1987) of the biceps femoris (ABF, 6 cats). For the ST both wires were


SCRATCHING

IN NORMAL

positioned proximal to the tendinous inscription, and for the LG electrode wires were secured within the lateral edge, a fast-twitch compartment (English 1984; Smith et al. 1977). In one additional cat (3.0 kg), a force transducer was implanted on the tendon of the medial gastrocnemius (MG), and EMG electrodes were implanted in MG and TA (see ACKNOWLEDGEMENTS). The transducer consisted of an oval-shaped, stainlesssteel buckle (16 mm long, 7.5 mm wide), a pair of foil strain gauges (MM 120 52)bonded to the oval top, and a center bar to secure the tendon (see Whiting et al. 1984 for design details). Gauges were connected to a Wheatstone bridge configuration with output voltages recorded on FM tape. Before surgery the transducer was calibrated statically by hanging known weights from a nylon cord passing through the buckle; gauge responses were linear from 0.5 to 120 N (Sherif et al. 1983). A final in vivo calibration was made during the terminal experiment (see cat B, Whiting et al. 1984).

Testing protocol Within 7- 10 days of surgery, testing began without discomfort to the cats and with no observable change in scratching behavior compared with that observed before surgery. During each recording session, a cat was placed in a 56 cm, Plexiglas testing cube; this enclosed environment limited the cat’s locomotion but not the scratching behavior. Responses were elicited by inserting a small ball of masking tape (2-3 mm) in the outer ear canal (Bradley and Smith 1988) or by rubbing a cotton swab on the inner wall of the pinna (called the concha by Sherrington 19 17) and outer ear canal. We limited stimulation to these areas because our preliminary observations and previous reports suggested that the receptive field is usually restricted to these areas (Abraham and Loeb 1985; Bradley and Smith 1988; Jankowska 1959; Sherrington 1917). At the beginning of each recording session, electrode placement was verified by muscle stimulation through the electrodes. Myopotentials were amplified (X 1,OOO), high-pass filtered (100 Hz), and recorded on FM tape (9.5 cm/s). The cat’s responses were recorded on video tape with a binary time code synchronized to the FM tape. For two cats, high-speed tine film ( 100 frames/s) was used to obtain data for the kinematic analyses of hindlimb scratching movements. Before filming circular markers (0.5 cm diam) were glued to the skin overlying the following bony landmarks: the pelvis (iliac crest), hip (greater trochanter), knee (lateral femoral epicondyle), ankle (lateral malleolus), and paw (fifth metatarsophalangeal joint, 5th MP). Scratch responses were filmed by a motor-driven, pin-registered, 16 mm tine camera that was positioned 2.5 m from the Plexiglas box with the optical axis perpendicular to the hindlimb. Film speed was verified by camera-internal timing lights actuated at 100 Hz. To synchronize the EMG and film records, a light-emitting diode placed in camera

CATS

1655

view was turned on and off repeatedly during filming, while a voltage pulse from the diode was recorded simultaneously on FM tape.

Data analysis Analog signals from each EMG channel were transferred to a digital code on an IBM PC/AT computer where the signals were sampled at 1 kHz, full-wave rectified, and stored on tape. A computer program (ANAPAC, Run Technologies, Los Angeles, CA) was used to detect onset and termination of each EMG burst when the signal increased and decreased, respectively, about a threshold voltage that ranged from 0.03 to 0.2 V, depending on average baseline noise. Because the TA exhibited the most consistent cyclic bursting (see RESULTS), the ankle flexor was selected as the reference muscle, and cycle period was defined as the time interval between onset of consecutive TA bursts. Onset latencies for other muscles were determined as the time between a TA burst onset and the burst onset of another muscle within a given cycle. Similarly, offset latencies were calculated as the time between TA burst onset and the burst offset of another muscle within a given cycle. Approximately 50 cycles from three to seven multicycle responses were analyzed for each animal, and responses selected were characterized by low-baseline noise and the highest number of rhythmically active muscles. From 18 responses recorded on tine film, 10 responses (2 cats) were selected for kinematic analysis based on the visibility of limb markers. Rectangular coordinates from consecutive film frames were digitized with the use of an overhead projection system and digitizer; data were stored on microcomputer disk for analysis. Only cycles in which limb motions were reasonably orthogonal to the optical axis of the tine camera were analyzed. Planarity was calculated from the difference between the directly measured thigh and shank lengths and the segment lengths calculated from film data; a maximum discrepancy of 10% was permitted. Data from most cycles of five responses from one cat satisfied this criteria, and for the second cat a maximum deviation of 12% was permitted to include cycles from one response. Kinematics illustrated in this paper are from the first cat but are representative of the data from both cats. Kinematic data were smoothed with the use of a cubic spline (Reinsch 1967). Paw contact was determined from film and synchronized to EMG via the light pulse. RESULTS

General description of the scratch response Scratching was often difficult to elicit because the response did not always follow stimulus application. After

FIG. 1. Postures adopted for the scratch response included lying (A), sitting (B), and standing (C). Each figure, drawn from a video image, shows the position of the hindlimb at paw contact. A and B are of one cat; C is of another-the same cat illustrated in Fig. 2.


1656

KUHTA

the stimulus was placed in the outer ear canal, the animal often ignored it and appeared easily distracted by environmental disturbances such as motions of the tester, movements of EMG cables, or the presence of a toy in the testing box. Also, the cats often employed movements other than scratching to dislodge the stimulus. Occasionally, they swiped at the stimulated ear with the ipsilateral forepaw, but most often head shaking followed stimulus application. Head-shake responses, described by Bradley and Smith (1988) and Sherrington ( 19 17), consisted of one or several rapid head rotations and was often effective in removing the stimulus. If a head shake or forepaw swipe failed to remove the stimulus, a scratch response sometimes followed. And, if the cat was unsuccessful in removing the stimulus by scratching, subsequent scratches were sometimes evoked by repositioning the stimulus with a cotton swab or by rubbing the base of the pinna. In preparation for scratching, the cats assumed one of three different postures (Fig. 1); these postures were not regulated by the tester, because we were interested in char-

(7 I / % \-r 7

AND

SMITH

A

]]]

5

4

2

3 i

6

1

f

floor

Approach

B

50 60

--f----

Cyclic ____t_

Return

1 1

I II,

0

250

I ,111

500

1111

750

I

1000

Time (ms) FIG. 3. Hindlimb kinematics for a standing scratch response with 3 contact cycles. A: each pair of stick figures represents a different period of the scratch response -approach (left), cyclic (ruliddle), and return (right). Dots on stick figures represent the iliac crest (top) and center positions for the hip, knee, ankle, and 5th MP joints. Arrows indicate direction of movement. Numbers (l-6) adjacent to stick figures also mark displacement records in B to identify the position of the hindlimb during each period of the response. B: joint-angle displacements for the knee (K) and ankle (A) joints. Arrows mark the reversal of knee motion during approach (left 4) and return (right 4) periods, and vertical lines mark the onset of contact for the first 3 cycles. At the end of the response, an attenuated cycle (e) resulted in no contact.

FIG. 2. Trajectories of the head and hindpaw during a 13-cycle scratch response. Head and paw trajectories are represented by locations of the nose (corner of right nostril) and 5th MP joint, respectively. During the approach period the head moved downward and caudally (- - - a) and the paw was lifted (- - - c). Nose position at paw off is marked by an arrowhead adjacent to line a. Dots indicated nose and paw positions at the onset of each contact phase; positions at contact for cycle 1 are marked by 1, whereas 3 and 7 indicate contact positions for cycles 3 and 7, respectively. Cat is shown in the contact position for cycle 6, and an open circle marks the 5th MP position. During the return period the head moved upward and rostrally (- - - b), and the paw returned to the floor (- - - d). Nose position at the moment the paw touched the floor is marked by an arrowhead adjacent to dashed line b, and onset of a head-shake response is indicated by a diamond (see text). Inset: nose area is expanded and changes in head position from the onset of cycle 6 to the onset of cycle 7 are shown. During the contact phase of cycle 6, the head moved rostrally (A, 10ms apart), and during the post- and precontact phases, the head moved clockwise (0, 10 ms apart).

acterizing normal posture during scratching. Most often the scratch response was initiated from a sitting position (Fig. 1B, 60% of responses); other responses were initiated from a standing position in which a tripod stance was adopted (Fig. lC, 26%) or from a lying position (Fig. IA, 14%). For six cats responses were recorded with the animal in at least two postures; for the other three cats responses were recorded with the animal in only one of the three postures. Although EMG data represent sitting, standing, and lying responses (see RESULTS, Muscle-activity patterns during the scratch response), kinematic data were analyzed primarily from standing responses because hindlimb motions during these responses usually met our planarity criteria (see METHODS), whereas those recorded with the cats in other postures did not. Some nonplanar motion was due to hindlimb action in the frontal plane, similar to that reported by Deliagina et al. (1975); however, most of the out-of-plane motion resulted from the cat sitting or standing with the scratching hindlimb oriented nonorthogonally to the camera lens. No lying scratches were filmed, because the hindlimb was positioned horizontal to the floor (Fig.


SCRATCHING

IN NORMAL

A

1657

CATS

B 50 -

5

4

110 y

6

/\ A H

K

Time

E 42

IOO6

;;; = z .-E 4 : 4

go-

80-

(set)

Ankle Joint Angle (deg)

coordination during cyclical scratching. A: angular displacements at the hip (H), knee (K), and ankle FIG. 4. Hindlimb (A) joint for cycles 4, 5, and 6 of a 13-cycle response (same response illustrated in Fig. 2). Vertical lines mark the onset and horizontal bars the duration of contact phases. B: angular motion at the ankle joint plotted against angular motion at the knee joint for same cycles. Shaded triangle marks the beginning of cycle 4 (at the onset of contact), and the plot follows a Unshaded triangle marks the termination of cycle 6 (e.g., last frame of the precontact phase). clockwise rotation ( c).

lA), and scratching motions occurred primarily in the frontal plane. Regardless of the posture adopted by the cat, the entire scratch response could be divided into three periods: approach, cyclic, and return. During the approach period the head and hindlimb ipsilateral to the stimulated pinna were positioned for scratching. The cyclic period was characterized by scratch cycles during which the hindpaw contacted

A

the head or pinna. Finally, in the return period the head and hindlimb resumed the preresponse posture. PERIOD. The approach period was initiated by head movements that positioned the stimulated pinna toward the ipsilateral hindlimb. When scratching was initiated from a lying position, the head was elevated above the trunk and rotated toward the hindlimb (Fig. 1A). From a

APPROACH

B

Pre

Contact

\\ \\ \\ \\ \,

.\

\\ \\

i/ i

100

1

\\

\\ \\ II II \ ’ 1 ‘./

I

Post ;A :i/ ,-\ \ \\ /:r \\ I' .; \\ /' / ;:: II ;K II III II III ;H /I : I III ~I :

A Time

FIG. 5. Hindlimb trajectory (A) and joint-angle excursions (B) for 2 different cycles from the same response. A: small dots illustrate movements of the iliac crest, hip and knee joints during a scratch cycle (same as cycle 6 in Fig. 4). Symbols at the ankle and 5th MP joints indicate positions during precontact (A), contact (A), and postcontact (0) phases; there are 10 ms between symbols. B: displacements for hip (H), knee (K), and ankle (A) joints are illustrated (same as cycle 5 of Fig. 4). Two solid vertical lines mark the onset and termination of the contact phase, whereas the dotted vertical line indicates the transition between post- and precontact phases. Time between large tick marks on the abscissa is 50 ms.


1658

KUHTA

sitting or standing position, in contrast, the head was lowered (Fig. 1, B and C), and as illustrated in Fig. 2, the head moved diagonally downward and caudally toward the hindlimb. Head movements appeared to result from a combination of ventral and lateral flexion of the neck; additionally, the neck rotated so that the stimulated pinna was lower than the contralateral ear (Figs. 1 and 2). Hindlimb movements, initiated after the onset of headpositioning motions, placed the paw near the base of the pinna during the approach period (Fig. 2). The paw trajectories for two typical standing responses are illustrated in Figs. 2 and 3A, l-2. At the beginning of the approach period, the upward paw trajectory resulted from flexion at the hip, knee, and ankle joints; however, knee flexion soon reversed to extension (4 left, Fig. 3B), and the paw approached the pinna by a combination of hip flexion, knee extension, and ankle flexion. The range of knee extension depended on the caudal-most position of the head as well as the degree of ankle and hip flexion, and typical hindlimb positions at the end of the approach period are illustrated in Figs. l-3. Collectively, these figures illustrate that regardless of the cat’s posture, hindlimb joints tended to be flexed to an angle ~90” at the moment of paw contact. CYCLIC PERIOD. A period of cyclical scratching immediately followed the approach period, and the number of scratch cycles per response varied from only a few cycles, such as the 3-cycle response illustrated in Fig. 3, to many cycles. The average number of cycles per response was 13 t

AND

SMITH

9, with the range from 1 to 60 cycles, and the cycle frequency ranged from 3.7 to 8 cycles/s. During most cycles the paw contacted the head, but for some cycles, particularly at the end of a response, the paw failed to make contact. In the response illustrated in Fig. 3, for example, the paw contacted the base of the pinna three times (marked by vertical lines, Fig. 3B); but after the third contact, there was an attenuated cycle (*) in which no contact occurred. During cyclical scratching there was usually angular displacement at the hip, knee, and ankle joints, and the range of motion at the ankle joint was typically greater than that at the knee (Figs. 3B and 4A) or hip (Fig. 4A) joints. The range of hip-joint motion was usually < 10”; but data on hip-joint motions were limited to a few responses, because the pelvic marker was often difficult to see in the film. Detailed descriptions of the hindlimb kinematics are given for exemplar scratch cycles in the following section (see RESULTS, Scratch-cycle kinematics). The head position often changed from cycle to cycle, and in Fig. 2 a plot of the head trajectory during a 13-cycle response illustrates this point. Here, the head moved caudally with successive contacts from cycles 1 to 4, but from cycles 5 to 7 the trajectory reversed, and the head moved rostrally with successive contacts. From cycles 8 to 13 the head was lowered but remained relatively stationary along the rostrocaudal axis. RETURN PERIOD. At the end of the cyclic period, the hindlimb and head returned to the preresponse posture,

*

LG

TA

. -._-.

--

B VL yr4u

_.--____-

I

FIG. 6. Muscle activity during the approach period and beginning of the cyclic period. A: first TA and IP bursts occurred during the approach period as the paw was positioned for scratching. Because the response was initiated from a standing position, the LG (an ankle extensor) was active (‘) before the limb was lifted from the ground. Approach-period activity was immediately followed by cyclic bursting, with alternate flexor and extensor activity. B: two extensor muscles (VL and LG) did not exhibit cyclical bursting at the beginning of the cyclic period (see RESULTS for further comments). Calibration bars: horizontal = 100 ms, vertical = 1 mV.

LG


SCRATCHING

IN NORMAL

and both trajectories were generally the reverse of the approach period, with the hindpaw returning to the floor before the head assumed the preresponse position (Fig. 2). As shown by hindlimb kinematics in Fig. 3A, 5-6, the paw trajectory during the return period was due to a combination of hip and ankle extension with knee flexion that reversed to knee extension (4, right of Fig. 3B). A head-shake response sometimes occurred during the return period but not until the elevated hindpaw contacted the floor, thereby providing a stable quadrupedal posture from which to initiate head shaking (cf. diamond adjacent to - - - b, Fig. 2). For the scratch response illustrated in Fig. 2, the head continued a diagonal-upward trajectory during the head-shake response as the cat resumed a standing position. Scratch-cycle kinematics

CATS

1659

lustrated for one cycle in Fig. 2 (inset). Here, the head moved rostrally in response to limb extension during the contact phase of cycle 6. The range of head excursion during the contact phase varied from cycle to cycle, and in some cycles head movement was extensive, and these cycles were usually associated with a more rounded paw trajectory during contact (i.e., Fig. 3A, 3-4). In other cycles, however, the head moved less, and these cycles were characterized by a flatter paw trajectory during the contact phase (i.e., Fig. 5A). POSTCONTACT AND PRECONTACT PHASES. During the postcontact phase the paw was withdrawn from the head by a combination of hip, knee, and ankle flexion, with ankle flexion dominating (Figs. 4A and 5B). In most cycles knee flexion occurred before ankle flexion at the end of the contact phase, and the transition from contact to postcontact is indicated by a clockwise loop in the angle-angle plots (top right loops, Fig. 4B). In some cycles onset of knee and ankle flexion was separated by lo-20 ms; this resulted in a narrow transition loop (e.g., cycles 5 and 6, Fig. 4B), but in other cycles knee flexion occurred 40-60 ms before ankle flexion, resulting in a broad transition loop (e.g., cycle 4, Fig. 4B). The onset of hip flexion followed the onset of ankle flexion, and in some cycles hip extension continued into the first part of the postcontact phase (i.e., Fig. 5B). The onset of the precontact phase was determined by the initiation of knee extension, and during this phase knee and ankle extension combined with hip flexion (Figs. 4A and 5B) to move the paw forward and upward (Fig. 5A). Knee extension always preceded ankle extension; thus the transition from the postcontact to the precontact phase is indicated by a clockwise loop in the angle-angle plots (bottom left loops, Fig. 4B). The hip joint continued to flex during the precontact phase, reaching peak flexion at or slightly after the onset of paw contact (Figs. 4A and 5B).

The circular path of the hindpaw trajectory during a single scratch cycle is illustrated in Fig. 5A (also see Fig. 3A, 3-4). The scratch cycle was divided into three phases: the contact phase during which the plantar surface of the paw contacted the head (A, Fig. 5A), the post-contact phase in which the paw moved away from the head (0, Fig. 5A), and the precontact phase in which the paw moved toward the head in preparation for another contact (A, Fig. 5A). From three different responses (1 cat) in which the phases could be measured accurately, the contact phase occupied 50 t 12% (20 cycles) of the scratch cycle, the postcontact phase occupied 24 t 9% (18 cycles), and the precontact phase occupied 26 t 9% (mean t SD; 21 cycles). CONTACT PHASE. Paw contact typically occurred near the stimulation site with the plantar pads contacting the upper neck, inside or behind the base of the pinna (Figs. 1 and 2); the exact contact location often varied from cycle to cycle and was not posture related. The hindpaw digits were usually flexed during contact, especially when the cat attempted to scratch the concha. Precise determination of the frames for paw onset and offset was often difficult as Muscle-activity patterns during the scratch response the plantar surface of the paw could be obscured by neck From 29 responses (7 cats) selected for EMG analyses, a fur or the pinna (e.g., - - - of the paw trajectory, Fig. 5A). total of 345 scratch cycles were analyzed. Of these cycles, Onset of paw contact was taken as the film frame where the paw appeared to make definite contact with the pinna or head, rather than merely brushing by the outer pinna. The 2001 0 -1 first film frame where the paw no longer touched the pinna or head was considered the end of the contact phase. The contact phase was characterized by extension at the hip, knee, and ankle joints, with ankle extension dominating (Fig. 5B). During initial cycles of some multicycle responses, the range of extension increased over successive cycles. This is illustrated for cycles 4-6 of a 13,cycle response by three angle-angle plots (Fig. 4B); here, ankle motion was plotted against knee motion, and peak extension progressively increased during three successive cycles I ’ I 1 I (top right loops, Fig. 4B). Although peak extension in100 150 200 250 300 creased, peak flexion at the end of the postcontact phase did not change (bottom left loops of Fig. 4B); consequently, Cycle Period (ms) knee and ankle joint positions at the onset of contact FIG. 7. Relationship between cycle period and burst duration for TA (marked by arrowheads in Fig. 4B) were similar for the and LG. Regression lines indicate that cycle period increased as burst three cycles. duration increased. Data included are from cycles where both TA and LG The head also tended to move during the contact phase, had bursting activity (n = 160 cycles). Circles indicate TA data, dots and a typical contact trajectory for head movement is il- indicate LG data. Both conelations were significant (P I 0.05).

‘m; g 150


1660

KUHTA

GM

Cycles 110 -117-1

ABF

121

I

I-lPJI-----+

VL

125

LG

-(I61

SOL EDL

121 47

TA

r 0

I

I

20

I

I

40 Percent

I

I

60 of Cycle

I

I

80

I

i

339

1 00

FIG. 8. Summary of muscle activities for cyclic scratching. For each muscle, the bar represents the average burst duration normalized to the TA cycle period. Standard deviation of onset and offset latencies for each burst duration are denoted by horizontal lines to the left and right of each bar, respectively. Flexor muscles are indicated by shaded bars and extensor muscles by unshaded bars. Number of cycles analyzed for each muscle is listed at the right; refer to METHODS for the number of cats tested for each muscle.

the number assessedfor each posture was proportional to the total number of cycles recorded for each posture: sitting (62%), lying (2 l%), and standing (17%). For all muscles except ST, preliminary analyses of the data revealed no significant posture- or cat-related differences; therefore the data were pooled for all cycles analyzed. A single, long burst of flexor activity characterized the approach period of the scratch response as the hindpaw was positioned toward the head. A typical example of the approach activity is illustrated for the IP and TA in Fig. 6A. The ankle extensor (LG), which had been active as the cat stood before the response (see *, Fig. 6A), was inactive during the approach as the cat assumed a tripod stance; this was also true of other extensor muscles except for the VL. In about one-half of the responses, the VL was active during the approach period, and this activity appeared to coincide with knee extension that accompanied hip and ankle flexion at the end of the approach period. Flexor activity, typical of the approach period, was usually followed by cyclic activity in which an alternating extensor and flexor pattern emerged (Fig. 6A). Usually, the TA exhibited repetitive bursts during the entire cyclic period; but other muscles, particularly the extensors, often exhibited low-level activity at the beginning of the cyclic period that gradually increased in amplitude and developed into distinct bursts (see VL and LG, Fig. 6B). Similarly, near the end of a scratch response, cyclical extensor bursts often ceased before those of the TA. Absence of repetitive bursting at the beginning of the cyclic period and early termination of such activity was common for extensor muscles (LG, VL, ABF, and GM), occurring in ~25% of the responses recorded. In some responses one or more extensor muscle exhibited low-level, tonic activity throughout the entire cyclic period, and in a few responses there was no EMG activity recorded throughout the response for the VL (3 responses), ABF (4), and GM (3). Cycle periods, defined as the interval between the onset of successive TA bursts, averaged 177 t 3 1 ms (339 cycles); thus the tvpical freauencv of repetitive TA bursting ranged

AND

SMITH

from 5 to 7 cycles/s. At the ankle, TA burst durations, with an average of 92 t 34 ms (52% of the cycle), were alternately active with LG and SOL. The burst durations for the two ankle extensors were similar, averaging 71 t 32 ms (40% of the cycle). The relationship between cycle period and burst duration was determined by linear regression, and as illustrated in Fig. 7, both TA and LG burst durations tended to increase with cycle period. The activity of the EDL, an ankle flexor and toe extensor, primarily overlapped activity of the TA, but on average EDL activity was shifted later in the scratch cycle than TA (Fig. 8). The activity of flexor and extensor muscles at the hip and knee joints tended to be paired with the activity of ankleflexor and -extensor muscles, respectively (Fig. 8). For example, IP activity was largely coactive with TA-EMG, but the IP burst usually preceded the TA burst. Likewise, extensor activity at the knee (VL) coincided with ankle-extensor activity, but on average VL activity preceded ankleextensor activity. In some cycles VL activity was shifted earlier, and in these cycles VL-EMG substantially overlapped TA-EMG. Extensor activity at the hip (GM and ABF) also coincided with ankle-extensor activity, but the onset of activity in hip extensor occurred later in the cycle (Fig. 8).

A ST

I

I

ST dd

I

TA

I

FIG. 9. Patterns of ST activity referenced to the TA. During cyclical scratching the temporal activity of the ST and TA was different for the 2 cats tested. A: ST was primarily coactive with extensors. B: ST more coactive with flexors (also see DISCUSSION). Calibration bars: horizontal = 100 ms. vertical = 0.25 mV.


SCRATCHING

IN NORMAL

The ST-EMG was recorded from two cats, and patterns of ST activity were sufficiently different (and animal related) that we were reluctant to average the data; thus samples of the two patterns are illustrated in Fig. 9. In one animal a long ST burst coincided with the TA interburst interval (Fig. 94, a pattern similar to that of the ankle extensors. In the other cat, however, the ST burst was rela-

CATS

1661

A SOL

A 65 L/L 75

v I

1

H

/Y

I

1

I

I

!

I

1

1

I

1

0

B 1 I

2

1

SOL

B

VI

IP

GM

VA

__

0 ABF

d

n

s P a

80

=E

A K

.0 7

D

TA FIG. 10. Two scratch cycles with joint kinematics synchronized to muscle activity (same as cycles 5 and 6 of Fig. 4). A: hip (H) joint excursions are matched with hip-muscle activity in B. C: knee (K) and ankle (A) joint angle excursions are paired with knee (VL) and ankle (TA, SOL) muscle activity in D. Vertical lines indicate the onset and horizontal bars the duration of contact. Calibration: vertical = 1 mV, horizontal = 50 ms between tick marks.

FIG. 1 1. Scratch responses evoked by a clip stimulus. A: cat scratched at a clip attached to the pinna while in a sitting posture. B: cat was in a lying posture. During both responses the paw lightly swiped at the clip, and the SOL exhibited brief bursts. Both A and B are averaged records triggered by the onset of TA activity (vertical line at 0). B: numbers 1 and 2 above the SOL-EMG indicate the sequence of the biphasic activity that centers around the TA burst (also see DISCUSSION). Calibration bars: horizontal = 50 ms between tick marks; vertical, SOL = 0.2 mV and TA = 0.1 mV.

tively short and primarily coincident with the beginning of a relatively long TA burst (Fig. 9B). Despite gross differences in the appearance of the two ST-EMG records, there are similar elements. In both, the ST burst was initiated during the TA interburst interval and was terminated during the first one-half of the TA burst; thus activity of this bifunctional muscle appeared to bridge extensor and flexor phases (also see DISCUSSION). The timing of muscle activity relative to joint kinematics and the contact phase is illustrated for two exemplar scratch cycles (Fig. 10). Paw contact was preceded by knee extensor (VL) and ankle extensor (SOL) activity, whereas hip-extensor activity (ABF, GM) occurred at or immediately after contact. The order of extensor recruitment was consistent with the sequence of joint reversals, with knee extension occurring before ankle and hip extension. During the paw-contact phase, joint extension was accompanied by extensor EMG. Flexor activity began near the end of the contact phase, and in this record IP and TA activity had similar onsets. In general, the onset of flexor activity preceded the initiation of joint flexion, and flexor activity


1662

KUHTA

AND

SMITH

coincided with joint flexion; similarly, extensor activity was initiated before peak joint flexion and coincided with joint extension. Novel variations of the scratch response STIMULUS-RELATED were able to elicit

MOTOR

PATTERNS.

In

one

cat

we

scratch responses by placing a small, straight-edged clip at the apex of the pinna or by placing a small tapeball in the outer ear canal, and a comparison of the EMG records revealed a stimulus-related motor pattern for the scratch cycle. During clip responses the cat swiped directly and precisely at the clip, making no head contact; this resulted in a brief contact phase and an equally brief bout of SOL-EMG (33 t 16 ms), occupying 19% of the cycle ( 178 t 13 ms; 53 cycles). During tapeball responses, in contrast, firm head contact occurred, and SOL activity typically lasted 68 t 30 ms, or 32% of the cycle (206 t 25 ms; 42 cycles). In some clip responses, SOL-EMG was biphasic with a primary burst before the TA burst (SOL-l, Fig. 11B) and a substantially smaller, second burst (SOL-2, Fig. 1lB) immediately after the TA burst. For responses initiated from a sitting position, the average TA burst duration was similar regardless of the stimulus, generally lasting -86 t 20 ms for responses elicited by clip (Fig. 1IA) and tapeball. However, when clip responses were elicited with the cat in a lying position, TA activity was markedly shorter and similar in duration (47 t 32 ms) to that of the SOL (Fig. 11B). These posture-related differences were specific to TA activity and not typical of SOL activity. EFFECTS OF IMPEDING THE CAT’S PAW. In one cat, MG tendon forces were recorded along with MG- and TAEMG during one scratch response (see acknowledgements). The response consisted of 19 cycles and was novel because of the variations in force during successive contact phases. To elicit scratching, the tester stimulated the concha with a cotton swab, and by chance, the tester’s hand was positioned between the cat’s hindpaw and the pinna; thus the paw’s normal trajectory was impeded, and the paw contacted the tester’s hand instead of the pinna. During the first 10 cycles, the paw lightly brushed the tester’s hand, but during the last nine cycles, the paw made firm contact, actually pushing the tester’s hand as the limb extended during contact. Peak tendon forces averaged 3.0 N for the light-contact cycles and 6.2 N for the firm-contact cycles (Fig. 12, A and B), and the cycle period increased from an average of 138 t 14 ms for the first 10 cycles to 18 1 t 20 ms for the last nine cycles. During light-contact cycles, the MG burst duration averaged 48 k 25 ms but nearly doubled to 84 ? 12 ms during firm-contact cycles. For both the light- and firmcontact cycles, the onset of MG activity preceded a rise in tendon force (marked by the vertical line in Fig. 12) and terminated at or just before peak tendon force. During the light-contact cycles the sharp-rising EMG burst was brief, but during the firm-contact cycles MG-EMG was biphasic, with a low-amplitude component briefly separated from a high-amplitude component. In contrast to changing MG activity, TA-EMG was relatively constant throughout the 19 cycles, averaging 46 t 11 ms for light-contact and 42 +

MG-F

TA

I

MG

I

.

MG-F ~...;--_b L

J

I

I

1

I

I

1

I

A

1

0 FIG. 12. Tendon forces during scratch cycles with light (A) and firm (B) paw contacts against the experimenter’s hand. See RESULTS for further description of this novel scratch response. Both records are averaged cycles triggered to the onset of MG tendon force (vertical line at 0). Time between horizontal ticks = 40 ms. Vertical calibration for muscles = 1 mV and force = 6.2 N.

11 ms for firm-contact cycles. Thus extensor and flexor bursts were similar in duration for light-contact cycles, but extensor bursts were twice as long as flexor bursts during firm-contact cycles. The absolute onset latency of MG recruitment was the same between light- and hard-contact cycles, but relative to the cycle period, MG was recruited late in the cycle (7 1%) for light-contact and midcycle (49%) for firm-contact cycles. DISCUSSION

Scratching in the adult cat was not easily evoked, and as Sherrington ( 1917) aptly remarked, “The scratch reaction often showed a very long latency, the cat seeming indisposed to permit it to ensue.” When we were able to evoke scratching in adult cats, the scratch response could be di-


SCRATCHING

IN NORMAL

vided into three periods (approach, cyclic, and return). During the approach period, also called the “postural” (Deliagina et al. 1975) or “initial aiming” (Baev 198 1) period, the hindlimb was positioned for scratching. In our experiments the cat’s head was also allowed to orient toward the hindpaw. The approach motions were accomplished with the cat in a lying, sitting, or standing position, and the preresponse posture was resumed during the return period after the termination of cyclic scratching. During the cyclic period the hindpaw usually contacted the head during each scratch cycle, and the typical cycle frequency (5-7 cycles/s) was similar to that for scratch responses of 3to 8-wk-old kittens (Bradley and Smith 1988) but faster than decerebrate air scratching, which is usually 3-4 cycles/s (Deliagina et al. 1975). Behavioral differences, such as cycle frequency, may be related to several factors, including the preparation, stimulation mode, and preresponse posturing. In the discussion that follows, we concentrate on how these factors and others influence limb kinematics and the relative timing of flexor and extensor activity during the scratch cycle. Scratch-cycle phases and limb kinematics We divided the scratch cycle into three phases (precontact, contact, postcontact), reminiscent of those introduced by Stein (1983; prerub, rub, and postrub) for the turtle scratch cycle. Our choice of terms denotes important behavioral differences between scratch responses of the turtle and cat. The turtle foot rubs across the stimulated site on an immovable shell bridge, and the rub phase occupies only -20% of the cycle (Mortin et al. 1985). The cat, in contrast, uses the plantar surface of the paw to contact the head, usually at the base of the pinna. Typically, the paw’s motion across the skin was minimal during contact; instead, the head moved rostrally as a result of hindlimb extension. Firm head contact may be needed to dislodge an ear-canal stimulus, and the laboratory-evoked behavior mimicked that normally used by cats to evict ear mites from the outer ear canal. The circular trajectories of the hindpaw during scratch cycles with head contact resembled those illustrated for air-scratch cycles in the decerebrate cat (Deliagina et al. 1975; Esipenko 1988). For air scratching, the trajectory and cycle are usually divided into a short phase (-30-50 ms) with the paw circling downward and a long phase (-250-300 ms) with the paw circling upward. The duration and timing of knee and ankle extension during the short-extensor phase of air scratching are similar to those of the precontact phase of the normal scratch cycle. And, although the flexor phase and limb kinematics of air scratching are similar to the postcontact phase of normal scratch cycle, the duration of limb flexion is considerably shorter for the normal- than the air-scratch cycle, 40-45 ms versus 250-300 ms, respectively. The primary difference between scratch cycles of the normal-behaving and decerebrate cats is that air scratching lacks the contact phase that occupies fully 50% of the normal-scratch cycle. During the contact phase the cat’s scratch cycle is characterized by extension at the hip, knee, and ankle joints. Although the turtle scratch cycle is characterized by knee

CATS

1663

extension during the rub, hip-joint motion depends on the scratch form (Mortin et al. 1985). For the rostra1 scratch, the turtle’s hip joint is held at peak flexion as the knee extends during the rub, but the hip extends during the rub phase of pocket and caudal scratches. By comparison, the cat scratch is similar to the turtle’s rostra1 scratch; in both, contact (or rub) is initiated with the hip joint at or near peak flexion (also see Bakker and Crowe 1982). During contact, however, the cat’s hip joint extended, whereas the turtle’s hip joint remained flexed. At the end of contact, the cat’s hindpaw was withdrawn from the head, first by the action of knee flexion and then by ankle flexion, followed last by hip flexion. Initially, then, the hip joint continued to extend as the knee flexed during the postcontact phase. The combination of hip extension and knee flexion is also typical of the postrub phase of the rostra1 turtle scratch. For the decerebrate air-scratch cycle, knee flexion also leads ankle and hip flexion, and the long-flexion phase is often characterized by a pause during which the ankle joint is held in flexion before the onset of extension (cf. Figs. 2A, 3, and 4 of Deliagina et al. 1975). Our kinematic data, in contrast, show a continuous flow of motion from post- to precontact phases; we observed no pause in limb motion after peak flexion. The precontact phase of the scratch cycle was initiated by knee extension, while the hip and ankle continued to flex, and this intralimb coordination is similar to that described by Esipenko (1988) for the “rhythmic aiming phase” of decerebrate air scratching (also see Baev 198 1). Esipenko (1988) divides the air-scratch cycle into an “aiming” and a “scratch” phase, and aiming occupies 80% (- 300 ms) of the air-scratch cycle, which is comparable in duration with the long-flexor phase described by Deliagina et al. (1975). The precontact phase of normal scratching, in contrast, was markedly shorter and occupied only 26% (40-55 ms) of the scratch cycle. Although the ankle was flexing at the beginning of the precontact phase, ankle motion reversed to combine with knee extension and hip flexion before paw contact. The combination of hip flexion and knee extension is also characteristic of the turtle’s leg during the prerub phase of the rostra1 scratch. Muscle synergies typical of the scratch cycle Regardless of the preparation used to study the cat scratch response, patterns of muscle activity are remarkably similar, and with few exceptions muscles can be grouped according to flexor and extensor functions, following Sherrington’s 19 10 classification. In the normal cat, extensor-muscle activity was generally confined to the precontact and contact phases. Although there was an interval during contact in which all extensor muscles tested were coactive, extensor recruitment was sequential: knee extensors were active before ankle extensors and ankle extensors before hip extensors. Typically, knee and ankle extensors were recruited before paw contact, whereas the onset of hip-extensor activity coincided with or occurred after the onset of contact. The knee-ankle-hip sequence of extensor recruitment was consistent with hindlimb kinematics; knee extension marked the onset of the precontact phase, followed by ankle extension and then hip extension.


1664

KUHTA

On average, we found VL activity coincided with ankle extensor activity, but in some cycles VL-EMG substantially overlapped TA-EMG. We suspect that an earlierthan-usual onset of VL activity was characteristic of cycles in which knee extension markedly preceded ankle extension during the precontact phase, but we lacked the kinematic records to confirm this. Baev (198 1) actually grouped uniarticular knee extensors with flexor muscles, and a summary of his data shows VL and TA neurograms were coactive during the fictive scratch cycle (see Fig. 3 of Baev 198 1). In discussing the action of the VL during the scratch response in decapitate cat, Sherrington ( 19 10) classified the VL with extensors of the hip and ankle. Also, Deliagina et al. (1975) listed the VL with other extensor muscles, showing that VL was usually active during the short-extensor phase and rarely active during the long-flexor phase (cf. Fig. 5 of Deliagina et al. 1975). In their study of scratch responses in kittens, Bradley and Smith ( 1988) found that VL recruitment usually preceded LG recruitment by 12-20 ms, but in one kitten a shorter-than-average VL burst was initiated 30-65 ms after the onset of the LG burst (see Fig. 1B of Bradley and Smith 1988). The late VL pattern was not observed in scratch records of adult cats; however, we noted scratch cycles in which VL activity was virtually absent. Flexor-muscle activity was generally confined to the postcontact phase, and generally, the onset of hip-flexor activity preceded ankle-flexor activity, a sequence typical of air and fictive scratching (Deliagina et al. 1975). Additionally, the bifunctional EDL (ankle flexor, toe extensor) was coactive with hip and ankle flexors, with EDL onset occurring after that of the TA. In our records the EDL burst overlapped the first part of the extensor burst, but unlike Deliagina et al. (1975), we did not observe two bouts of EDL activity during a single scratch cycle. From our EMG data it was difficult to classify ST activity as part of the flexor or extensor synergy. Sherrington (19 10) noted that the ST and its anatomic synergist, the posterior biceps femoris (PBF), were coactive with flexor muscles during the scratch reflex. Conversely, Baev (198 1) and Deliagina et al. ( 1975) grouped ST and PBF activity as part of the short-extensor phase. Later, in a study of motoneuron activity during fictive scratching, Berkinblit et al. (1980) showed that PBF and ST (or collectively PBST) motoneurons exhibited two peaks of depolarization per cycle-one at the middle of the short-extensor phase and one at the beginning of the long-flexor phase. A similar pattern of depolarization was described for PBST-Ia inhibitory neurons (Deliagina and Orlovsky 1980). Our ST-EMG data are consistent with these findings, as ST activity was initiated during the middle of the extensor phase and continued into the beginning of the flexor phase. In this manner, ST activity spanned the extensor-flexor transition from the contact to postcontact phase. Although the motor pattern for scratching is usually characterized by reciprocal pattern of flexor and extensor activity (Gelfand et al. 1988), our EMG records, those of Bradley and Smith (1988), and the electroneurographic (ENG) records of others (Baev 198 1; Berkinblit 1978a,b; Deliagina et al. 1975) show periods of cocontraction. Gen-

AND

SMITH

erally, the onset of extensor activity overlapped the offset of flexor activity and vice versa. Coactivity of this type is common for other cyclical hindlimb motions that are traditionally classified as having reciprocal flexor-extensor patterns. During the step cycle, for example, TA and LG activity overlap during the El phase as the paw is lowered for stance (Buford and Smith 1990; Engberg and Lundberg 1969). However, in addition to periods of cocontraction during the scratch cycle, there were also responses in which one or more muscles had low-level tonic activity for several cycles or the entire response. Tonic activity in some muscles may relate to the maintenance of limb attitude during the scratch cycle, as activity of this nature is not typical of walking (Grillner 198 l), airstepping (Giuliani and Smith 1985), or paw shaking (Smith et al. 1985). Relative timing of’eor

and extensor bursts

Although muscle synergies during the scratch cycle are similar for normal and decerebrate cats, the relative timing of flexor and extensor activity is markedly different. Notably, decerebrate air scratching is characterized by flexor activity that is 8- 10 times longer than extensor activity (Baev 198 1; Berkinblit et al. 1978a,b, 1980; Deliagina et al. 1975, 198 1, 1983). The asymmetrical flexor-extensor pattern is not typical of scratching in adult cats (reported here) (also see Abraham and Loeb 1985) or kittens (Bradley and Smith 1988). In 1980 Berkinblit et al. reported a few fictive scratch responses in which flexor and extensor phases were equal in duration, and noted that “such efferent activity is typical for stepping.” Fast walk and trot step cycles are characterized by reciprocal extensor and flexor activity of equal duration (Engberg and Lundberg 1969; Grillner 198 1), but this symmetrical pattern is also characteristic of most scratch cycles in the normal cat. Our records show that extensor EMG preceded paw contact by 20-40 ms, and interestingly, this is similar to the duration of extensor activity for air- and fictive-scratch cycles in which no contact phase occurs (Baev 198 1; Berkinblit et al. 1978a,b, 1980; Deliagina et al. 1975, 198 1, 1983). Possibly, hindlimb feedback during contact-either cutaneous, proprioceptive, or both-prolongs extensor activity and delays the onset of flexor activity to make the pattern of activity more symmetrical. Prolongation of extensor activity was illustrated in a novel record of the rostral turtle scratch (Stein and Grossman 1980). During one cycle the claws accidentally caught the stimulating probe and the foot pushed against it, increasing both the amplitude and duration of hip-extensor activity and delaying the onset of hip-flexor activity (see Fig. 10 of Stein and Grossman 1980). The role of feedback in prolonging extensor activity during paw contact has been investigated for the cat’s step cycle by Duysens and Pearson (1980); specifically, they found that when triceps surae tendon forces exceeded 4 kg (-39 N) during stance, extensor activity was prolonged, and the onset of flexor activity was inhibited. It is unlikely that triceps surae tendon forces peak to this level during the contact phase of scratching, because large-extensor contractile forces would lead to large contact forces at the head. Initially, then, extensor activity may be facilitated during


SCRATCHING

IN NORMAL

contact, but as contractile tension increases, feedback may act to inhibit extensor activity to prevent the development of large contact forces during scratching. We found, for example, that the MG tendon force increased during contact, with the onset of TA activity occurring about the time of the peak tendon force. These preliminary data suggest that receptors monitoring increases in contractile force, such as the tendon organ (see review by Stuart et al. 197 1; but also see Powers and Binder 1985), may be responsible for inhibiting extensor activity and facilitating flexor activity to terminate the contact phase. We do not know the magnitudes of peak tendon or contact forces that occur when the cat’s paw makes head contact during normal scratching, and we are in the process of collecting these data. We found, however, that when the scratching paw made firm contact with the experimenter’s hand, MG tendon forces peak at -6 N (see Fig. 12B); this level is nearly 70% of the peak MG tendon force that develops during stance of slow walking (see Table 4, cat B of Whiting et al. 1984) and similar to that for quiet standing (Walmsley et al. 1978). That extensor tendon forces are lower for contact during scratching than walking is consistent with the different demands of the two behaviors, and feedback regulation may be different for the two movements. In addition to input from hindlimb muscle and joint receptors, feedback from cutaneous receptors of the paw and contact site (pinna or neck) may effect the timing of extensor- and flexor-muscle activity. Also, input from neck muscle proprioceptors may modulate the motor program via connections with descending propriospinal interneurons. The cat’s head is not stationary during the contact phase of the scratch cycle, and the discharge of muscle spindle afferents may reflect changes in neck motion, and muscle spindles, in particular, are abundant in the dorsal neck muscles (Richmond and Abrahams 1975; Richmond et al. 1986). The role that neck proprioception plays in modulating hindlimb motor patterns or activities of neck muscles during the scratch cycle is unknown. The complex interaction between hindlimb and head motions during contact has not been studied and warrants consideration in future studies of the scratch response. The influence of proprioception or cutaneous input on the scratch motor pattern has been given little attention in studies of decerebrate scratching. Some information, though, is available on the effects of hindlimb position on motor activities during fictive scratching. Baev ( 198 1; Shimanskii and Baev 1987), for example, reported that drawing the limb forward by hip flexion increased the duration and amplitude of extensor activity but decreased flexor activity. They also showed that drawing the limb backward by hip extension reduced extensor activity, and, if the limb was drawn back far enough, extensor activity was markedly reduced or absent altogether. The latter finding is consistent with earlier observations by Deliagina et al. (1975) that decerebrate cat scratching could not be evoked with the hip joint extended. If, however, the decerebrate cat’s hindlimb is deafferented, fictive scratching is evoked independent of hip-joint position (Berkinblit et al. 1978b). Although Deliagina et al. (1975) concluded that limb deafferentation resulted only in a change in muscle-activity

CATS

1665

levels during decerebrate air scratching, Esipenko ( 1988) found that deafferentation had “a considerable effect on the parameters and general pattern of firing in the muscles.” In particular, ankle-extensor activity (LG) was markedly shorter (see Fig. 6d of Esipenko 1988), and the duration of ankle extension was considerably shortened after deafferentation. Also, Shimanskii ( 1988) demonstrated that passive hindlimb motions, mimicking the air-scratch trajectory, were able to entrain the cyclic activity of fictive scratching in the decerebrate, immobilized cat. These studies suggest that although lumbosacral circuits generate scratch-like patterns, the input from phasic and tonic limb afference has the capacity to modify the motor output. Modulation by motion-related feedback is only one way in which extensor activity may be prolonged during the scratch cycle. Extensor activity is also modulated by tonic facilitator-y input from propriospinal fibers that presumably trigger scratching. Berkinblit et al. (1980), for example, showed that extensor motoneurons may depolarize without an action potential at the beginning of a fictive scratch record, but as the “intensity of scratching increased,” the depolarization and number of action potentials increased from two to six (see Fig. 3, C and D of Berkinblit et al. 1980). Similar modulatory effects on flexor motoneurons were not observed during fictive scratching. Input from some descending tracts may also increase the duration of extensor activity during the scratch cycle. During the approach period, for example, the cat’s head is tilted with the stimulated ear lower than the nonstimulated ear, and the head tilt is maintained throughout the period of cyclic scratching. Arshavsky et al. (1979) showed that an ipsilateral head tilt resulted in an increase in the cyclic bursting (i.e., longer bursts and higher firing rates) of ipsilateral vestibulospinal (Deiter’s) neurons during fictive scratching in decerebrate cat. Because vestibulospinal neurons with excitatory input to extensor hindlimb motoneurons are maximally active during the extensor phase of the scratch cycle (Arshavsky et al. 1978), it is probable that head tilting-typical of the behaving cat-serves to augment the duration and intensity of extensor activity programed by the spinal generator. Thoughts on forms of scratching and pattern generation Although scratching is usually classified as a stereotypic hindlimb behavior, our data suggest that the response in normal cats is variable and not particularly stereotypic. In a study of three cyclical behaviors of the cat’s hindlimb (walking, paw shaking, and scratching), Abraham and Loeb (1985) noted that “scratching was the most variable of the activities studied, with animals exhibiting individual styles of limb and claw posture.” We agree with this observation and their suggestion that response variations are due in part to the cat’s posture. For the most part, however, variations are related to differences in the onset and offset of muscles within the two synergies, flexor versus extensor, and from our data we were unable to identify reliable posture-related differences in motor patterns for ear-canalevoked scratches. We observed one posture-related difference in the timing of the flexor bursts for clip responses; the TA activity was markedly longer during sitting than lying


1666

KUHTA

AND

responses, and this may be related to the position of the hindlimb relative to gravity. In the future systematic studies may reveal other posture-dependent patterns. Another parameter that might account for motor-pattern variability is the location of the scratch stimulus, but because the receptive field for evoking the cat scratch response is limited primarily to the pinna, neck, and upper shoulder areas (Bradley and Smith 1988; Sherrington 19 lo), it is unlikely that distinct behavioral styles, such as the three forms of the turtle scratch (Mortin et al. 1985; Robertson et al. 1985) or multiple forms of the frog wiping reflex (Berkinblit et al. 1986; Giszter et al. 1989), will characterize the cat scratch response. But the type of stimulus used to evoke the cat scratch response may result in different styles of scratching. The clip stimulus, for example, elicited a response in which the paw precisely and lightly brushed tip of the pinna, and extensor-flexor burst patterns were reminiscent of rapid aiming movements in which the first agonist burst (SOL- 1 burst, Fig. 11B) was followed closely by an antagonistic burst (TA burst) and a second agonist burst (SOL-2, Fig. 1lB), the “three-burst pattern” defined by Ghez and Martin (1982; also see Ghez and Gordon 1987). Other types of scratching, such as removing parasites (fleas) from the skin, may elicit more of a vigorous brushing action across the skin rather than a relatively stationary and firm head contact. Data from extensive research on fictive scratching (Baev 198 1; Berkinblit 1978a,b, 1980; Deliagina et al. 1975, 198 1, 1983; Shimanskii 1988; Shimanskii and Baev 1987) are consistent with the hypothesis that lumbosacral circuits generate a scratch-like motor program without phasic supraspinal influence or motion-related feedback. Whether the fictive scratch program represents the natural output of the spinal generator or one heavily biased by the mode of stimulation and/or supraspinal input unique to the decerebrate preparation remains to be determined. Although synergistic muscle patterns for scratching are similar in the normal and decerebrate cat, the relative timing of extensor and flexor activity is markedly different. Taken together these data suggest that the central pattern for scratching is only part of a more complex system that determines the relative timing of muscle activity. Bassler (1986) refers to the “pattern generator” (PG) as the combination of central (CPG) and peripheral activity and notes that there may be differences in the functional principles of pattern generation between CPG and PG. Pearson ( 1987) makes a similar argument, insisting that “To conclude that the CPG provides the timing of motor activity, it is necessary to show that the fictive pattern is similar to the intact motor pattern.” During the scratch cycle cutaneous and proprioceptive feedback may be critical to programing the relative timing of extensor and flexor activity. Specifically, we hypothesize that feedback during the contact phase of the scratch cycle is an integral part of the pattern-generating network for responses in normal cats. Additionally, the cat’s posture and particularly the attitude of the head may alter the activity of neurons from various descending tracts, including the vestibulospinal, as well as the rubrospinal and reticulospinal. Here, the cerebellum may play a critical role, as cyclic activity of these three descending tracts is absent

SMITH

after cerebellectomy in the decerebrate cat (see Arshavsky et al. 1986). What role various cortical areas or subcortical nuclei, such as the basal ganglia, play in selecting the posture and initiating the response is unknown. Future studies will need to concentrate on what type of feedback is essential, and how sensory input interacts with the spinal CPG as well as inputs from supraspinal centers to establish the appropriate timing of motor activity. We thank R. F. Zemicke for assistance in collecting the kinematic data and J. A. Buford for assistance with analysis of kinematic data and surgery. Also, we are grateful to V. R. Edger-ton for the loan of a cat with EMG electrodes implanted in the EDL [research supported by National Institute of Neurological Disorders and Stroke (NINDS) Grant NS-163331 and to R. J. Gregor and W. C. Whiting for the use of an FM tape (EMG and force records) that included a scratch record from one cat with an MG tendon force transducer (research supported by NINDS Grant NS- 16333). Locomotion data for this animal had been analyzed as part of another study (see cat B, Whiting et al. 1984); however, the scratch response was not analyzed. Our research is supported by NINDS Grant NS- 19864 and a Biomedical Research support grant from UCLA. Address for reprint requests: J. L. Smith, Dept. of Kinesiology, 2859 Slichter Hall, University of California, Los Angeles, CA 90024- 1568. Received 7 March 1990; accepted in final form 13 August 1990. REFERENCES L. D. AND LOEB, G. E. The distal hindlimb musculature ofthe cat. Patterns of normal use. Exp. Brain Res. 58: 580-593, 1985. ARSHAVSKY, Yu. I., GELFAND, I. M., AND ORLOVSKY, G. N. Cerebellum and Rhythmical Movements. New York: Springer-Verlag, 1986, p. 1-166. ARSHAVSKY, Yu. I., GELFAND, I. M., ORLOVSKY, G. N., AND PAVLOVA, G. A. Messages conveyed by descending tracts during scratching in the cat. I. Activity of vestibulospinal neurons. Brain Res. 159: 99-l 10, 1978. ARSHAVSKY, Yu. I., ORLOVSKY, G. N., AND PANCHIN, Y. V. Responses of Deiter’s neurons to tilt during scratching. Neurophysiol. Kiev 10: 229-231, 1979. BAEV, K. D. The central program of activation of hind-limb muscles during scratching. Neurophysiol. Kiev 13: 38-44, 198 1. BAEV, K. D., DEGTYARENKO, A. M., ZAVADSKAYA, T. V., AND KosTYUK, P. G. Activity of lumbosacral interneurons during fictitious scratching. Neurophysiol. Kiev 13: 45-52, 198 1. BAKKER, J. G. M. AND CROWE, A. Multicyclic scratch reflex movements in the terrapin Pseudemys scripta elegans. J. Comp. Physiol. 145: 477-484, 1982. B;~SSLER, U. On the definition of central pattern generator and its sensory control. Biol. Cybern. 54: 65-69, 1986. BERIUNBLIT, M. B., DELIAGINA, T. G., FELDMAN, A. G., GELFAND, I. M., AND ORLOVSKY, G. N. Generation of scratching. I. Activity of spinal interneurons during scratching. J. Neurophysiol. 4 1: 1040- 1057, 1978a. BERKINBLIT, M. B., DELIAGINA, T. G., FELDMAN, A. G., GELFAND, I. M., AND ORLOVSKY, G. N. Generation of scratching. II. Nonregular regimes of generation. J. Neurophysiol. 4 1: 1058- 1069, 1978b. BERKINBLIT, M. B., DELIAGINA, T. G., ORLOVSKY, G. N., AND FELDMAN, A. G. Activity of motoneurons during fictitious scratch reflex in the cat. Brain Res. 193: 427-438, 1980. BERKINBLIT, M. B., FELDMAN, A. G., AND FUKSON, 0. I. Adaptability of innate motor patterns and motor control mechanisms. Behav. Brain Sci. 9: 585-638, 1986. BETTS, B., SMITH, J. L., EDGERTON, V. R., AND COLLATOS, T.C.Telemetered EMG of fast and slow *muscles in cats. Brain Res. 117: 529-533, 1976. BRADLEY, N. S. AND SMITH, J. L. Neuromuscular patterns of stereotypic hindlimb behaviors in the first two postnatal months. III. Scratching and the paw-shake response in kittens. Dev. Brain Res. 38: 69-82, 1988. BUFORD, J. A. AND SMITH, J. L. Adaptive control for backward quadrupedal walking. II. Hindlimb muscle synergies. J. Neurophysiol. 64: 756-766, 1990. ABRAHAM,


SCRATCHING

IN NORMAL

P. AND SMITH, J. L. Characteristics of scratching in normal cats. Sot. Neurosci. Abstr. 14: 26 1, 1988. CARLSON KUHTA, P., SMITH, J. L., AND ZERNICKE, R. F. Hindlimb kinematics and muscle synergies during scratching. Sot. Neurosci. Abstr. 15: 394, 1989. DELIAGINA, T. G. AND FELDMAN, A. G. Activity of Renshaw cells during fictive scratch reflex in the cat. Exp. Brain Res. 42: 108- 115, 198 1. DELIAGINA, T. G., FELDMAN, A. G., GELFAND, I. M., AND ORLOVSKY, G. N. On the role of central pattern and afferent inflow in the control of scratching movements in the cat. Brain Res. 100: 297-3 13, 1975. DELIAGINA, T. G. AND ORLOVSKY, G. N. Activity of Ia interneurons during fictitious scratch reflex in the cat. Brain Res. 193: 439-447, 1980. DELIAGINA, T.G., ORLOVSKY,G.N., AND PAVLOVA, G.A.Thecapacity for generation of rhythmic oscillations is distributed in the lumbosacral spinal cord of the cat. Exp. Brain Res. 53: 8 l-90, 1983. DELIAGINA, T. G., ORLOVSKY, G. N., AND PERRET, C. Efferent activity during fictitious scratch reflex in the cat. J. Neurophysiol. 45: 595-604, 1981. D~MER, F. R. AND FELDBERG, W. Scratching movements and facilitation of the scratch reflex produced by tubocurarine in cats. J. Physiol. Land. 153: 35-51, 1960. DUYSENS, J. AND PEARSON, K. G. Inhibition of flexor burst generation by loading ankle extensor muscles in walking cats. Brain Res. 187: 321-332, 1980. ENGBERG, I. AND LUNDBERG, A. An electromyographic analysis of muscular activity in the hindlimb of the cat during unrestrained locomotion. Acta Physiol. Stand. 75: 6 14-630, 1969. ENGLISH, A. W. An electromyographic analysis of compartments in cat lateral gastrocnemius muscle during unrestrained locomotion. J. Neurophysiol. 52: 114- 125, 1984. ENGLISH, A. W. AND WEEKS, 0. I. An anatomical and functional analysis of cat biceps femoris and semitendinosus muscles. J. MorphoZ. 19 1: 161-175, 1987. ESIPENKO, V. B. Correlation between the kinematics of hindlimb movement and efferent activity in the decerebrate cat during scratching. Neurophysiol. Kiev 19: 398-405, 1988. FELDBERG, W. AND FLEISCHHAUER, K. Scratching movements evoked by drugs applied to the upper cervical cord. J. Physiol. Lond. 15 1: 502-5 17, 1960. FELDMAN, A. G., ORLOVSKY, G. N., AND PERRET, C. Activity of muscle spindle afferents during fictitious scratch reflex in the cat. Brain Res. 129: 192-196, 1977. GELFAND, I. M., ORLOVSKY, G. N., AND SHIK, M. L. Locomotion and scratching in tetrapods. In: Neural Control of Rhythmic Movements in Vertebrates, edited by A. H. Cohen, S. Rossignol, and S. Grillner. New York: Wiley, 1988, p. 167-199. GHEZ, C. AND GORDON, J. Trajectory control in target force impulses. I. Role of opposing muscles. Exp. Brain Res. 67: 225-240, 1987. GHEZ, C. AND MARTIN, J. H. The control of rapid limb movements in the cat. III. Agonist-antagonist coupling. Exp. Brain Res. 45: 115-125, 1982. GISZTER, S. F., MCINTYRE, J., AND BIZZI, E. Kinematic strategies and sensorimotor transformations in the wiping movements of frogs. J. Neurophysiol. 62: 750-767, 1989. GIULIANI, C. A. AND SMITH, J. L. Development of characteristics of airstepping in chronic spinal cats. J. Neurosci. 5: 1276-1282, 1985. GOLDBERGER, M. E. AND MURRAY, M. Restitution of function and collateral sprouting in the cat spinal cord: the deafferented animal. J. Camp. NeuroZ. 158: 37-54, 1974. GRAHAM BROWN, T. Studies in the reflexes of the guinea pig. I. The scratch-reflex in relation to “Brown-S&quard’s epilepsy.” Q. J. Exp. Physiol. 2: 243-275, 1909. GRILLNER, S. Control of locomotion in bipeds, tetrapods and fish. In: CARLSON KUHTA,

CATS

1667

Handbook of Physiology. The Nervous System. Motor Control. Washington, DC: Am. Physiol. Sot., 198 1, sect. 1, vol. II, p. 1179-l 236. JANKOWSKA, E. Instrumental scratch reflex of the deafferented limb in cats and rats. Acta BioZ. Exp. 19: 233-247, 1959. MORTIN, L. I., KEIFER, J., AND STEIN, P. S. G. Three forms of the scratch reflex in the spinal turtle: movement analysis. J. Neurophysiol. 53: 1501-1516, 1985. O'DONOVAN, M.J., PINTER, M.J., DUM, R.P., AND BURKE, R.E. Action of FDL and FHL muscles in intact cats: functional dissociation between anatomical synergists. J. Neurophysiol. 47: 1126-l 143, 1982. PEARSON, K. G. Central pattern generation: a concept under scrutiny. In: Advances in Physiological Research, edited by H. M. Lennan, J. R. Ledsome, C. H. S. McIntosh, and D. R. Jones. New York: Plenum, 1987, p. 167-185. POWERS, R. K. AND BINDER, M. D. Distribution of oligosynaptic group I input to the cat medial gastrocnemius motoneuron pool. J. Neurophysiol. 53: 497-5 17, 1985. REINSCH, C. H. Smoothing by spline functions. Numer. Math. 10: 177-183, 1967. RICHMOND, F. J. R. AND ABRAHAMS, V. C. Morphology and distribution of muscle spindles in dorsal muscles of the cat neck. J. Neurophysiol. 38: 1322-1339, 1975. RICHMOND, F.J.R., BAKKER,G. J., BAKKER, D.A., AND STACEY, M.J. The innervation of tandem muscle spindles in the cat neck. J. Comp. Neural. 245: 483-497, 1986. ROBERTSON,G. A., MORTIN, L.I., KEIFER, J., AND STEIN, P.S.G.Three forms of the scratch reflex in the spinal turtle: central generation of motor patterns. J. Neurophysiol. 53: 15 17-l 534, 1985. SHERIF, M. H., GREGOR, R. J., LUI, L. M., ROY, R. R., AND HAGER, C. L. Correlation of myoelectric activity and muscle force during selected cat treadmill locomotion. J. Biomech. 16: 69 l-701, 1983. SHERRINGTON, C. S. Observations on the scratch-reflex in the spinal dog. J. Physiol. Lond. 34: l-50, 1906. SHERRINGTON, C. S. Notes on the scratch-reflex of the cat. Q. J. Exp. Physiol. 3: 213-220, 1910. SHERRINGTON, C. S. Reflexes elicitable in the cat from pinna, vibrissae, and jaws. J. Physiol. Lond. 5 1: 40 l-43 1, 19 17. SHIMANSKII, Yu. P. Recording of scratch generator efferent activity produced by cyclic hindlimb movement in decerebrate immobilized cat. Neurophysiol. Kiev 19: 324-328, 1988. SHIMANSKII, Yu. P. AND BAEV, K. V. Dependence of efferent activity parameters on limb position during fictitious scratching in decerebrate cats. Neurophysiol. Kiev 15: 45 l-458, 1987. SMITH, J.L., EDGERTON, V.R., BETTS, B., AND COLLATOS, T.C.EMGof slow and fast extensors of cat during posture, locomotion, and jumping. J. Neurophysiol. 40: 503-5 13, 1977. SMITH, J. L., HOY, M. G., KOSHLAND, G. F., PHILLIPS, D. M., AND ZERNICKE, R. F. Intralimb coordination of the paw-shake response: a novel mixed synergy. J. Neurophysiol. 54: 127 l- 128 1, 1985. STEIN, P. S. G. The vertebrate scratch reflex. In: Neural Origin of Rhythmic Movements, edited by A. Roberts and B. Roberts. New York: Cambridge Univ. Press, 1983, p. 383-403. STEIN, P. S. G. AND GROSSMAN, M. L. Central program for scratch reflex in turtle. J. Comp. Physiol. 140: 287-294, 1980. STUART, D. G., MOSHER, C. G., AND GERLACH, R. L. Properties and central connections of Golgi tendon organs with specific reference to locomotion. In: Research in Muscle DeveZopment in the Muscle Spindle. Excerpta Medica International Congress Series, p. 437-476, 197 1. WALMSLEY, B., HODGSON, J. A., AND BURKE, R.E.Forcesproducedby medial gastrocnemius and soleus muscles during locomotion in freely moving cats. J. Neurophysiol. 41: 1203-1216, 1978. WHITING, W.C., GREGOR, R.J., ROY, R.R., AND EDGERTON, V.R. A technique for estimating mechanical work of individual muscles in the cat during treadmill locomotion. J. Biomech. 17: 685-694, 1984.


Kinematics and control of frog hindlimb movements.

Responses of spinal cord neurons to systematic changes in hindlimb skin temperatures in cats and primates.

Dense distributed processing in a hindlimb scratch motor network.

Development of coordinated movement in chicks: I. Temporal analysis of hindlimb muscle synergies at embryonic days 9 and 10.

Shared and Task-Specific Muscle Synergies during Normal Walking and Slipping.

Locomotor recovery after unilateral hindlimb deafferentation in cats.

Weight-bearing hindlimb stepping in treadmill-exercised adult spinal cats.

Pharmacokinetics of methimazole in normal cats and cats with hyperthyroidism.

Empirical Evaluation of Voluntarily Activatable Muscle Synergies.

Kinematics and muscular activity of cats during maximum height jumps.

Development of coordinated movement in chicks: II. Temporal analysis of hindlimb muscle synergies at embryonic day 10 in embryos with spinal gap transections.

Brain Connectivity Associated with Muscle Synergies in Humans.

Muscle fibre number following hindlimb immobilization.

Muscle synergies: input or output variables for neural control?

Muscle Synergies Facilitate Computational Prediction of Subject-Specific Walking Motions.

Interactions between locomotor and respiratory activities during hindlimb stepping on a stationary surface in decerebrate cats.

Ventral horn cell responses to spaceflight and hindlimb suspension.

Electrical and mechanical responses in the platysma and in the adductor pollicis muscle: in normal subjects.

Muscle synergies evoked by microstimulation are preferentially encoded during behavior.

Regenerative responses in cultured hindlimb stumps of larval Xenopus laevis.

Reorganization of muscle synergies during multidirectional reaching in the horizontal plane with experimental muscle pain.

Forelimb and hindlimb ground reaction forces of walking cats: assessment and comparison with walking dogs.

Erythrocyte protoporphyrin concentrations in clinically normal cats and cats with lead toxicity.

Development of visuomotor behavior in normal and dark-reared cats.