COWAN, N.; LEE, J.; FULL, R.J.; Johns Hopkins Univ.; Univ. of California, Berkeley: Dynamical control of antennal wall following in the American cockroach

During rapid locomotion, the antenna of a cockroach, Periplaneta americana , transduces contact and strain stimuli to neural impulses. These impulses are used for locomotion control, enabling the cockroach to execute rapid maneuvers of up to 25 turns/sec (Camhi and Johnson 1999). Using data of several cockroaches running on a flat surface following angled walls (30 to 60 deg), we determined a third order dynamical control model for wall following behavior. In the model, a cockroach antenna measures distance and rate of approach to the wall, from which the cockroach generates a rotational moment to change its orientation. To fit the parameters of the model, we measured the response time of a cockroach’s body angle as it encountered an angled wall while running at speeds from 25 – 60 cm/sec. We found that the time to first achieve the new angle (rise time) decreased linearly from 0.35s to 0.2s with an increase in speed. The time to reach and maintain steady state at the final angle (settling time ~0.5s) was independent of speed. These findings confirm the proposed dynamical control system model. To determine how the antenna might encode distance, we investigated how antennal posture affects a cockroach’s ability to follow a wall. When the antenna projected forward along the wall, the cockroach ran much closer (13.8 � 3.6mm) to it than when the antenna was bent backward (20.6 � 5.2mm). For a given antenna posture, contact area and strain both increase with decreased cockroach-to-wall distance. However, on the projected-forward trials, there was significantly more contact and less strain than the bent-backward trials. Therefore, based on our control model, contact area alone is not used for control, while it is possible that strain alone is sufficient. Supported by ONR/DARPA. N00014-98-1-0747.