(Received 10 April 1956)

446 J. Physiol. (I956) I33, 446-455 A COMPARISON OF FLEXOR AND EXTENSOR REFLEXES OF MUSCULAR ORIGIN BY M. G. F. FUORTES AND D. H. HUBEL From the Department ofneurophysiology, Walter Reed Army Institute ofresearch, Walter Reed Army Medical Center, Washington 12, D.C. (Received 10 April 1956) Whereas extensor muscles of decerebrate mammalian preparations usually present both phasic and tonic reflex contraction to stretch (Liddell & Sherrington, 1925), flexor muscles present phasic stretch reflexes only (Asayama, 1916). Under anaesthesia or during spinal shock (Creed, Denny-Brown, Eccles, Liddell & Sherrington, 1932, p. 51), tonic stretch reflexes are lost while phasic responses are maintained. Extensor and flexor muscles then respond in a similar way to applied stretch. The present research was undertaken with the purpose of identifying the processes responsible for this difference between flexor and extensor pathways. This appeared to be a convenient way of approaching an understanding of the central events which underlie the development of sustained reflex responses. METHODS The experiments were performed on thirty-two acutely decerebrate cats. Decerebration was carried out under deep ether anaesthesia and additional anaesthetic was administered during the subsequent surgical procedure if required. In all cases the experiments were performed after recovery, as judged by restoration of such anaesthetic-sensitive reflex movements as flexor withdrawal, extensor thrust, contralateral extensor reflex, etc. The lumbar enlargement of the spinal cord was exposed and appropriate ventral roots were severed on one side near the point of passage through the dura. Records were taken from L7 or S1 ventral roots or from one of their filaments, with neighbouring ventral roots cut to prevent antidromic invasion. The branches of the ipsilateral hamstring and gastrocnemius nerves were severed and their proximal stumps were used for stimulation. Stimulation and recording were carried out in the conventional way in most experiments, but in some a train of stimuli was used for conditioning and a subsequent single shock for testing. This was done by using the first pulse of a doublechannel stimulator (Attree, 1950) as a gate for another stimulator supplying the train of conditioning shocks, and the second pulse as a test stimulus (Fig. 1). RESULTS Responses to repetitive stimulation In previous work (Alvord & Fuortes, 1953, fig. 8) it was found that in decerebrate preparations repetitive stimuli to extensor muscle afferent fibres

FLEXOR AND EXTENSOR REFLEXES 447 insufficient to evoke reflex firing of extensor motoneurones when delivered at a low frequency might evoke sustained firing if applied at higher frequencies. It was noted that the firing evoked in this way was similar to that elicited in the same motoneurones by sustained stretching of the muscle. Fig. 1. Block diagram of stimulation arrangement when trains of conditioning stimuli were used. Stimulator A provides two impulses separated by a variable delay: the first of these serves as a gate for a single-channel stimulator B which provides a train of stimuli used for conditioning; the second provides the test stimulus. Below is shown a record led off from a small filament of SI ventral root. A train of eight stimuli is used for conditioning one branch of the hamstring nerve. The following test shock is applied to the second branch. As illustrated in Fig. 2, different features are typical of flexor muscular responses. A flexor motoneurone responds to single-shock stimulation of a flexor muscular nerve with the discharge of a single impulse occurring after monosynaptic delay (Lloyd, 1943; Alvord & Fuortes, 1954). With weak repetitive shocks applied at a low frequency (e.g. 0.5/sec-1/sec) a response will occur to some of the stimuli. Increase of stimulus strength causes an increase in the ratio of number of reflex impulses to number of stimuli (probability of firing) until eventually the motoneurone responds to each stimulus. If frequency of stimulation is now increased (its strength being unchanged) this ratio decreases. For a moderate increase of frequency (say from 1/sec to 2/sec) responses occur to some of the stimuli throughout the period of stimulation. At higher frequencies (10/sec-100/sec) the motoneurone responds only to the first stimulus or to the first few stimuli, and usually no further impulses break through even if the stimulation is maintained for a long time (Fuortes, 1954 c). While the greater effectiveness of the first stimulus is particularly obvious for higher frequencies of stimulation, it is also often observed for slow stimulation rates such as 0.5-1/sec, where the first of the stimuli usually evokes 29 PHYSIO. CXXXIII

448 M. G. F. FUORTES AND D. H. HUBEL a response even if probability of firing is quite low for the stimuli following the first. If frequency of stimulation is set at a high value (e.g. 100/sec) and the voltage is gradually increased from zero, no firing at all results even if the final voltage reaches or exceeds values which are effective in producing a response when the stimulation is applied abruptly (Fig. 3). Fig. 2. Reflex firing of a single motoneurone in response to stimuli to one branch of the hamstring nerve. Records taken from a thin filament of a peripherally cut ventral root. Strength of stimulation constant throughout; frequency 1, 10, 100/sec as indicated by artifacts. Time 1 sec. These results are analogous to the findings obtained with flexor reflexes evoked by stretch by Asayama (1916), who observed that flexor muscles react with a short-lasting contraction to a pluck or to an abrupt stretch but do not respond to a gradual increase in tension. As noted in previous work (see, for example, Fuortes, 1954 b), sustained responses are sensitive to many types of damage. Because of this it was considered important to show that the unsustained character of flexor muscular reflexes is not the consequence of the general factors which indiscriminately depress sustained responses. Fig. 4 illustrates the results obtained in one of the experiments done for this purpose. A ventral root filament containing axons of extensor and flexor units was selected for recording. A flexor motoneurone contained in this filament responded to stimulation of one branch of the hamstring nerve in the usual manner: it responded to all shocks if these were delivered at 1/sec, but only to the first if stimulated at 10/sec or 100/sec. In contrast, extensor units in the same filament failed to

FLEXOR AND EXTENSOR REFLEXES 449 respond to stimuli applied to one branch of the gastrocnemius nerve at 1/sec or 10/sec but readily responded if the frequency was raised to 100/sec. In other experiments (Fig. 5) it was observed that flexor motoneurones which respond only initially to repetitive stimulation of muscular nerves may present sustained firing on stimulation of skin nerves or receptors. A..W B Pm m.. m~in~ Fig. 3 Fig. 4 Fig. 3. Repetitive stimulation at 250/sec of one branch of hamstring nerve (lower beam), record taken from a small filament of S 1 ventral root (upper beam). A, Stimulation applied abruptly; B, intensity of stimulation gradually increased to a final value equalling that of A. Time 1 sec. Fig. 4. Responses of flexor (A) and extensor (B) motoneurones to repetitive stimulation of hamstring and gastrocnemius nerves respectively: all records taken within a short time from a ventral root filament containing one unit responding to hamstring and a few units responding to gastrocnemius nerve stimulation. Stimuli at 1, 10 and 100/sec indicated by artifacts. Time 1 sec. A Fig. 5. Reflex firing of a motoneurone recorded from a ventral root filament and evoked by stimulation of one branch of hamstring nerve at 10/sec in A and by pinching the skin of the ipsilateral hindlimb in B. Time 1 sec. 29-2

450 M. G. F. FUORTES AND D. H. HUBEL These results show that the early failure of flexor reflexes of muscular origin is not due to inherent inability ofthe preparation to produce sustained responses, as is the case in anaesthetized preparations; and also, that it is not a necessary feature of the flexor motoneurone. Rather, abrupt and unsustained responses are a property of flexor muscular reflex pathways. The different responses of flexor and extensor muscular pathways to repetitive stimulation show that volleys from flexor nerves contribute a condition which tends to prevent firing, whereas volleys from extensor nerves result in accumulation of a change which tends to facilitate a reflex response. Excitability changes after one or more stimuli In contrast with this conclusion derived from experiments employing repetitive stimulation, previous results obtained by testing excitability changes evoked on a motoneurone pool by a single impinging volley (Lloyd, 1946) have shown no differences between extensor and flexor pathways. This finding was not confirmed in preparations such as the unanaesthetized decerebrate, presenting unsustained reflexes of flexor but well-sustained reflexes of extensor muscles. The excitability cycle following a subliminal conditioning shock was determined in two ways: by recording from a large number of ventral root fibres and measuring the height of the reflex spike resulting from the synchronous firing of many units ('composite recording'), and by recording from a single motoneurone, determining its probability of firing in response to a given test shock at various times after conditioning. Typical results obtained with the first method are shown in Fig. 6. If a conditioning shock subliminal for reflex activation is applied to a flexor muscular nerve, the reflex spike elicited by a subsequent larger test shock to the same or to another closely allied nerve is increased for some 15 msec, but later is considerably decreased for a period which may greatly exceed 1 sec. In extensor muscular reflex pathways the early facilitation may be more prolonged but the subsequent depression is absent, or at least less pronounced. This is only true in preparations presenting well-sustained extensor reflexes: if these are unsustained (as happens under anaesthesia, spinal shock, surgical damage, or under other circumstances escaping experimental control), then strong depression is observed also in extensors. The difference in properties of flexor and extensor muscular reflexes is emphasized if a train of stimuli rather than a single stimulus is used for conditioning (Fig. 1). With trains of 5 to 12 shocks delivered in a period of 50-75 msec, it was observed that in flexor reflex pathways facilitation reaches a maximum after the first, second, or sometimes third stimulus, and then decreases sharply. Conversely, facilitation builds up progressively in extensor

FLEXOR AND EXTENSOR REFLEXES 451 pathways so that the effectiveness of a test stimulus is increased maximally towards the end of the train. This is illustrated with composite recording in Fig. 7. As with single conditioning shocks, the difference between flexors and extensors was evident only if extensor reflexes were well-sustained in experiments employing composite recording. Even in only moderately rigid preparations, however, a conspicuous difference was found between flexor and A I 0 B C D 0 50 100 msec D 50 100 msec Fig. 6. Amplitude of reflex spike recorded from a whole ventral root and evoked by stimuli applied at different times after a subliminal conditioning shock. The horizontal line in each record indicates the average height of response to test shock alone. Conditioning and test stimuli applied to the same branch of the hamstring nerve in A and separate branches in C; to the same branch of the gastrocnemius nerve in B, and separate branches in D. extensor units selected for typical unsustained activity on the one hand and for typical sustained activity on the other. Examples of the course of probability of firing of such units during and following conditioning with a train of impulses are given in Fig. 8. Fig. 9 shows the course of probability of firing following trains of different lengths, illustrating its decrease with increasing train length for flexor pathways, as contrasted with its increase in the case of extensors. Consistent with these results, it was found that when the strength of the train of conditioning stimuli was raised until it barely succeeded by itself in

452 M. G. F. FUORTES AND D. H. HUBEL exciting a flexor motoneurone, this discharged after one of the first three stimuli (see Eccles & Rall, 1951), while extensor motoneurones discharged after the last stimulus (Fig. 10). A 08 0 0 01* A 1229 24 1410 13 1110 21 10 3 i10 Il-* t- j I.i I I 0 _i1..a_ L,~ I6 i10 2 30 40 50 60 70 msec 0 10 20 30 40 50 60 msec B 15 15 B 15 15 15 1515 15 15 15 171515 8 1 00 co 00 0 0 0 0 5 0. 0 10 20 30 40 50 60 70 msec 0 10. 20 3 40 50 60 msec Fig. 7 Fig. 8 Fig. 7. Amplitude of reflex spike recorded from a whole ventral root and evoked by stimuli applied at different times during and after a subliminal (conditioning) train of shocks, indicated by the vertical lines. The horizontal line in each record represents the average height of response to test shock alone. Conditioning and test stimuli applied to the same branch of the hamstring nerve in A and of the gastrocnemius nerve in B. Fig. 8. Probability of firing of a single motoneurone in response to a test shock applied during and after a subliminal (conditioning) train of stimuli. Probability of firing is expressed as the ratio of number of responses obtained to number of stimuli at each time interval. The number of applied stimuli at a given interval is indicated by the figure above each point. Probability of firing in response to a test shock alone is indicated by the horizontal line in A, and was zero in B. A, Conditioning and testing stimuli applied to one branch of the hamstring nerve; B, same preparation as A, stimuli to one branch of the gastrocnemius nerve. DISCUSSION Prior to this study identical changes have been described as occurring in flexor and extensor muscular reflex arcs following their activation. It is important to note that those properties of muscular reflexes which are identical in flexor and extensor pathways cannot include features relevant to production of rhythmical firing, since rhythmical responses can be obtained in extensor pathways only. Therefore care should be exerted in applying conclusions derived from the study of so-called monosynaptic reflexes (see

A FLEXOR AND EXTENSOR REFLEXES 453 Alvord & Fuortes, 1953, p. 303) to the interpretation of the events occurring during the normal movements brought about by repetitive firing of each activated motoneurone. _0 o.... A.5 F a 3 _ b 3 0 E 0 5 0... a"00 _- 0 20 40 60 msec Time, 10 msec Fig. 9 Fig. lo Fig. 9. Same experimental arrangement as Fig. 8. Testing stimulus delivered after the end of conditioning trains of stimuli at 300/sec, the numbers besiide the curves indicating the number of impulses in a conditioning train. A, Stimuli to one branch of the hamstring nerve; B, same preparation as A: stimuli to one branch of gastrocnemius nerve. Probability of firing in response to a test shock alone was one in A and zero in B. Fig. lo. Impulses recorded from a thin ventral root filament of a decerebrate cat in response to a train of stimuli adjusited to be just threshold for reflex discharge. In A stimuli were applied to one branch of the hamstring nerve, and in B to one branch of the gastrocnemius. According to the results obtained in the present research, different features of temporal summation characterize flexor and exrtensor reflexres of muscular origin. Pathways capable of responding with rhythmical firing present exrtensive abilities for temporal summation. In these pathways, the facilitatory effect exrerted by repetitive impingement builds up to a steady level which, in the absence of motoneurone firing, can be maintained for a long time. In some conditions, the build-up of facilitation can progressively increase over a considerable fraction of a second, explaining the long latency between start of stimulation and the first reflex response observed on occasion in previous and in the present work (see Alvord & Fuortes, 1953, figs. 6, 7A and 8).

454 M. G. F. FUORTES AND D. H. HUBEL In pathways responding with typically unsustained discharge to repetitive stimulation, excitability reached a peak within a few msec and decayed thereafter, even in the absence of a reflex response. Consistently, reflex firing could be elicited in these pathways after short latencies only. The present work does not reveal the nature and location of the processes responsible for the contrasting features of flexor and extensor reflexes. In a previous study on the effects of direct currents on flexor and extensor motoneurones (Fuortes, 1954a), some difference was found between the two types of cells, but this would be insufficient to justify the sharp difference between their responses to repetitive afferent stimulation. Some other property of primary afferent terminals or of the central links activated by muscular stimulation in the two cases is likely to play a more important role. Even without commitment as to the cause of the different features of flexor and extensor reflex responses, it can be stated that the observed correlations between time-course of facilitation and ability to respond with rhythmical firing justify the conclusion that temporal summation of a change fitting Sherrington's definition of central excitatory state (Creed et al. 1932, pp. 31, 32) is necessary for production of rhythmical reflexes. SUMMARY 1. Reflex responses of flexor motoneurones to stimulation of hamstring nerve have been compared with responses of extensor motoneurones to stimulation of gastrocnemius nerve in decerebrate cats. 2. Whereas firing of extensor motoneurones is often more easily evoked by high than by low frequency of stimulation, flexor motoneurones responding to each shock when stimulation frequency is low respond only initially to stimuli delivered at higher frequencies. 3. Following a single conditioning shock to the hamstring nerve, the reflex response to a subsequent test shock is increased for a period of about 15 msec, but later is considerably decreased. In extensor pathways, the early facilitation may be longer and the late depression is either less pronounced or altogether absent. 4. With a train of conditioning stimuli, facilitation increases to an early maximum and soon decays in flexor pathways, but it builds up more or less gradually to a steady state in extensor pathways. 5. It is thought that the properties of temporal summation found in extensor pathways are necessary for production of sustained rhythmical reflex responses.

FLEXOR AND EXTENSOR REFLEXES 455 REFERENCES ALVORD, E. C. JR. & FUORTES, M. G. F. (1953). Reflex activity of extensor motor units following muscular afferent excitation. J. Physiol. 122, 302-321. ALVORD, E. C. JR. & FUORTES, M. G. F. (1954). A comparison of flexor reflexes of cutaneous and muscular origin. J. Physiol. 123, 251-259. ASAYAMA, C. (1916). The proprioceptive reflex of a flexor muscle. Quart. J. exp. Phy8iol. 9, 265-279. ATTREE, V. H. (1950). An electronic stimulator for biological research. Rev. 8ci. Instrum. 27, 43-47. CREED, R. S., DENNY-BROWN, D., ECCLES, J. C., LIDDELL, E. G. T. & SHERRrNGTON, C. S. (1932). Reflex Activity of the Spinal Cord. Oxford: Clarendon Press. ECCLES, J. C. & RATi, W. (1951). Repetitive monosynaptic activation of motoneurones. Proc. Roy. Soc. B, 138, 475-498. FUORTES, M. G. F. (1954a). Direct current stimulation of motoneurones. J. Physiol. 126, 494-506. FUORTES, M. G. F. (1954b). Activity of single motoneurones during some reflex reactions of mammals. C.R. 5th int. neurol. Congr. (Lisbon), 4, 36-51. FUORTES, M. G. F. (1954c). Interaction between flexor reflexes of cutaneous and of muscular origin. Amer. J. Physiol. 170, 636-637. LIDDELL, E. G. T. & SHERRINGTON, C. S. (1925). Reflexes in response to stretch (myotatic reflexes). Proc. Roy. Soc. B. 96, 212-242. LLOYD, D. P. C. (1943). Neuron patterns controlling transmission of ipsilateral hind limb reflexes in cat. J. Neurophysiol. 6, 293-315. LLOYD, D. P. C. (1946). Integrative pattern of excitation and inhibition in two-neuron reflex arcs. J. Neurophysiol. 9, 439-444.