The escape behavior of the cockroach is a ballistic behavior with well char
acterized kinematics. The circuitry known to control the behavior lies in t
he thoracic ganglia, abdominal ganglia, and abdominal nerve cord. Some evid
ence suggests inputs may occur from the brain or suboesophageal ganglion. W
e tested this notion by decapitating cockroaches, removing all descending i
nputs, and evoking escape responses. The decapitated cockroaches exhibited
directionally appropriate escape turns. However, there was a front-to-back
gradient of change: the front legs moved little if at all, the middle legs
moved in the proper direction but with reduced excursion, and the rear legs
moved normally. The same pattern was seen when only inputs from the brain
were removed, the suboesophageal ganglion remaining intact and connected to
the thoracic ganglia. Electromyogram (EMG) analysis showed that the loss o
f or reduction in excursion was accompanied by a loss of or reduction in fa
st motor neuron activity. The loss of fast motor neuron activity was also o
bserved in a reduced preparation in which descending neural signals were re
versibly blocked via an isotonic sucrose solution superfusing the neck conn
ectives, indicating that the changes seen were not due to trauma. Our data
demonstrate that while the thoracic circuitry is sufficient to produce dire
ctional escape, lesion or blockage of the connective affects the excitabili
ty of components of the escape circuitry. Because of the rapidity of the es
cape response, such effects are likely due to the elimination of tonic desc
ending inputs. (C) 2001 John Wiley & Sons, Inc.