The resolution of target range by echolocating bats

TLDR

Bats appear to use a neural matched‑filter that cross‑correlates emitted signals with returning echoes to determine echo arrival time. Range discrimination is achieved by timing differences in echo arrivals, with Eptesicus and Phyllostomus using the full FM sweep, Pteronotus employing its short CF/FM burst, and Rhinolophus extracting the FM portion of its long CF/FM call for ranging. All four species can resolve range differences as small as 1–3 cm, with the first three species showing range‑independent acuity at short distances, and the data support distinct roles for CF and FM components and validate a matched‑filter cross‑correlation strategy for bat echolocation.

Abstract

Using simultaneous discrimination procedures the acuity of resolution of differences in target range has been determined on four species of echolocating bats (Eptesicus fuscus, Phyllostomus hastatus, Pteronotus suapurensis, and Rhinolophus ferrumequinum). All can discriminate range differences as small as 1 to 3 cm and, for the first three species, the acuity of range resolution appears to be independent of absolute range, at least at short distances. In Eptesicus range discrimination mediated in terms of the arrival times of echoes returning from different targets. Comparisons between discrimination performance and autocorrelation functions of echolocation sounds used in the discriminations suggest that these bats possess some neural equivalent of a matched-filter, ideal sonar receiver which functionally cross-correlates a replica of the outgoing signal with the returning echo to detect the echo and determine its arrival time. Eptesicus and Phyllostomus both use the entire FM signal for target ranging. Pteronotus uses its entire compound, short-duration CF/FM signal for ranging, while Rhinolophus separates the FM component from its compound, long-duration CF/FM sound and uses the FM part for target ranging. The results indicate different functions for the CF and FM components of bat sonar cries, and they suggest that the matched-filter or cross-correlation approach to echolocation is appropriate.