Target detection and range resolution by the big brown bat (Eptesicus fuscus) using normal and time-reversed model echoes

Summary

Big brown bats (Eptesicus fuscus) were tested for their ability to detect an electronically simulated target, and to discriminate differences in range to two simulated targets, when receiving either a model of their own sonar emissions or the model reversed in time as the ‘echo’. The theory of matched detection predicts a large decrease in performance if bats use matched filtering, unless they are somehow able to adjust their filter to match the novel, time-reversed signal. The detection thresholds we obtained were much lower than Møhl's (1986), but like him we found no difference in threshold for reversed models (Table 2). This suggests either that bats do not use matched filtering for target detection, or, possibly, that they are able to adapt their filter to a highly unnatural signal in some way as yet unknown.

Unlike detection, range discrimination was much poorer with reversed echoes (Table 3). Threshold increased from about 1 cm range difference with normal model echoes to 18 cm or more with reversed model echoes. This suggests that range determination, which is based on measuring the time of arrival of echoes, does involve matched filtering. Whether such filtering is ‘ideal’ (i.e., equivalent to cross-correlation detection) cannot be decided by our results, but there are some indications that the match between an echo and the presumed internal template (the ‘match’ of matched filtering) must be fairly precise. Also, since performance with phantom targets generated using model echoes was as good as has been found with real targets, the internal template is probably fixed (or only slowly modifiable) rather than re-programmed with each sonar emission. Finally, because synchronization of emission and model echo was not perfect, the apparent distance to targets probably varied by 2 to 4 cm from emission to emission, although both targets would appear to move together thus keeping the range difference constant. This suggests that bats determined range to the targets simultaneously rather than sequentially, as is usually assumed.