172 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. I37 



is no longer maintained in the extended position. When sensory 

 input becomes inadequate, stimulation to the muscles holding the 

 proboscis in the extended position ceases, these muscles relax, and 

 the proboscis is passively partially retracted. By elastic recoil the 

 labellar lobes close, the haustellum tends to relax against the rostrum, 

 while the rostrum itself no longer remains fully extended. Failure of 

 adequate sensory input may result from adaptation of the chemo- 

 receptors themselves, central adaptation, or inhibition from the 

 stomatogastric nervous system. Dethier and Bodenstein (1958) have 

 shown that satiation in the fly can be equated with carbohydrate 

 in the foregut, that stimulation of unknown receptors there results in 

 impulses passing up the recurrent nerve to the brain where the effects 

 of otherwise adequate sensory input from the proboscis are inhibited. 



Sucking may cease as a direct result of impulses from rejection 

 neurons in the labellar hairs, the interpseudotracheal papillae, or the 

 tarsal hairs. Under these circumstances impulses must inhibit cen- 

 trally the initiation of potentials in the efferents to the dilators of the 

 cibarial pump. 



In contrast to the slow relaxation of the proboscis commonly oc- 

 curring when sensory input fails, the proboscis can be retracted ac- 

 tively with great speed when adverse stimuli are presented to the 

 labellar or tarsal chemoreceptors. Under these circumstances impulses 

 are routed to five sets of retractors and flexors as already described. 

 Present evidence seems to suggest that active retraction partakes 

 more of an all-or-none reaction than does extension ; however, fur- 

 ther work may reveal that here also there is recruitment of muscles 

 depending upon the intensity of the adverse stimulus. 



The situation becomes even more complicated when the fly is pre- 

 sented with mixtures of acceptable and unacceptable compounds. The 

 mixtures may be applied to a single chemoreceptive hair containing 

 acceptance and rejection neurons, or an acceptable compound may 

 be presented to a neuron on the tarsi simultaneously with presentation 

 of an unacceptable compound on a labellar neuron. In either case, the 

 resultant response depends on the nature of the balance obtained in 

 sensory input (cf. Dethier, 1955 ; Dethier, Evans, and Rhoades, 1956). 

 If sensory input from the acceptance neuron predominates, the 

 proboscis is extended; if sensory input from the rejection neuron 

 predominates, the proboscis is retracted. It is not surprising that a 

 balance involving such widely separated neurons as labellar and tarsal 

 may operate almost as effectively as a balance involving two labellar 

 hairs because, as has been shown, the input from these various sources 



