PREY CAPTURE IN MANTIDS 61 



constant correlation to the deviation of the prey from the body axis ; and 

 this again means that each of them — at equilibrium — contains the informa- 

 tion which is required to determine the correct direction of the stroke.^ 



Let us first assume that the direction of the stroke be determined by 

 the optic center message only (Fig. 6c, without the dotted arrow). It may 

 be useful to illustrate this by an example. For the sake of simplicity the 

 direction of the stroke k may be defined as the angle between the endpoint 

 of the stroke and the median plane of the prothorax, the axis of reference 

 being the vertical axis of the head movement. Then the prey will be hit 

 if K is equal to o-. Now consider the system has a fixation-deficit of \0% ; 

 that is — at equilibrium — the optic input equals -f 1 ° if the prey deviates 

 + 10° from the body axis, +2° if the prey deviates +20°, and so on. Evi- 

 dently the central units involved in the determination of the stroke merely 

 have to amplify the optic input by a constant factor in order to hit the prey 

 exactly. 



Let us see now how this hypothesis fits the experimental results first 

 presented. 



( 1 ) If the hair plates are eliminated by nerve resection on both sides, 

 the proprioceptive circuit breaks down. Thus the optic circuit normally 

 working against it will now be more effective in minimizing (f). Conse- 

 quently there must be a smaller fixation-deficit than before. Because the 

 factor by which the optic input is multiplied in order to steer the direction 

 of the stroke is presumably not changed by the operation, the deviations 

 of the stroke from the body axis will always be too small, except in the one 

 case when the prey is sitting straight ahead. As we have learned, this is 

 just what happens in fact. 



(2) If the head is given a fixed position, say 20° to the left, then the two 

 circuits are both blocked. Thus the performance will depend upon whether 

 the animal succeeds in centering the fly by means of movements of the legs 

 and the prothorax. If the fly is at least approximately centered that way, 

 before the stroke is released, the optic input will be about zero, the stroke 

 thus going more or less straight ahead. But since the fly then in fact has a 

 mean deviation of 20° to the left, it normally will be missed to the right, 

 quite in accordance with the facts. Yet it will be hit frequently, if the head 

 is fastened in its median position. As we have learned, that is indeed so. 



(3) The result of the combination experiment does not agree with the 

 assumption that the stroke is determined by the optic-center message only. 

 If the head cannot be moved, the additional elimination of proprioceptors 

 should then have no additional effect at all. Consequently we must conclude 



1 To satisfy the rigorist, it may be added that this is true even if we leave out the 

 linearity and proportionaHty conditions introduced at the beginning. 



