MECHANORECEPTORS AND BEHAVIOR 373 



value of the sharklike fishes in this area of research has lain in their rare 

 capacity for perfonning regular swimming movements after complete re- 

 moval of the brain. Consequently, and in complete contrast to other verte- 

 brates, the elasmobranch spinal preparation is one whose locomotory move- 

 ments are coordinated solely by spinal cord neurons. 



Since persistence of movement in spinal dogfish was first observed at the 

 end of the last century several workers have examined this preparation. The 

 work of Gray and Sand (1936), among others, was important in showing 

 that spinal neurons are capable of generating rhythmical activity even in the 

 absence of proprioceptive input, although Lissmann's de-afferentation ex- 

 periment (Lissmann 1946b) implied some role for proprioceptive feedback. 



These experiments have contributed to arguments as to whether loco- 

 motion in vertebrates is governed by "central rhythms," generated by "oscil- 

 lators" or produced by "chain reflexes," established by proprioceptive feed- 

 back. This debate, which is reminiscent of the polarised views seen in other 

 disciplines (such as the preformation-epigenesis controversy of embryology 

 and the nature-nuture debate of geneticists), in retrospect can be seen to 

 have been rather sterile and notable for lack of definition. Indeed, in view of 

 current ideas of corollary discharges and efferent supply of sense organs, the 

 distinction between "central" and "peripheral" becomes very blurred, and 

 the two extreme viewpoints are seen to be untenable. No movement could 

 be completely determined by the sensory input, which would make the 

 nervous system redundant, any more than the central nervous system could 

 be totally independent. A more illuminating approach to the role of sensory 

 activity in movement control is to attempt to define the activity of specific 

 sense organs during movement and to determine the functional value of this 

 activity. 



All the evidence so far for the sharklike elasmobranchs indicates that the 

 amplitudes and frequencies of the swimming movements are probably regu- 

 lated by two separate but interrelated mechanisms. Also, whereas the sen- 

 sory input, in one form or another, is important in sustaining the amplitude, 

 the frequency is determined more by the properties of central neurons, 

 which, depending on their excitability, tend to discharge spontaneously. 

 Recordings taken from the spinal nerves of curarized spinal fish have shown 

 that pronounced rhythmical discharges are obtained from the motor nerves 

 (Roberts 1969a, Grillner, Perret, and Zangger 1976). These discharges are 

 sustained even in the absence of proprioceptive activity and can be restarted 

 after they have waned by the application of nonphasic tactile stimulation 

 (Figure 28). These data indicate, not surprisingly, that much of the pattern 

 of movement is determined to a large extent by the spinal cord. Nevertheless, 

 changes in the timing of the proprioceptive feedback indicate that this too 

 must have some role, for when the body of a free-swimming spinal dogfish is 

 subjected to forced oscillation, the electromyographic records from the 

 swimming musculature show very clearly that the motor pattern becomes 

 immediately entrained to the applied rhythm (Figure 29). 



These two pieces of evidence can be incorporated in a general view of 

 proprioceptive function. This requires only the assumption that the tendency 



