Perceptual strategies in active and passive hearing of neotropical bats

Basic spectral and temporal sound properties, such as frequency
content and timing, are evaluated by the auditory system to build
an internal representation of the external world and to generate
auditory guided behaviour. Using echolocating bats as model system,
I investigated aspects of spectral and temporal processing during
echolocation and in relation to passive listening, and the
echo-acoustic object recognition for navigation. In the first
project (chapter 2), the spectral processing during passive and
active hearing was compared in the echolocting bat Phyllostomus
discolor. Sounds are ubiquitously used for many vital behaviours,
such as communication, predator and prey detection, or
echolocation. The frequency content of a sound is one major
component for the correct perception of the transmitted
information, but it is distorted while travelling from the sound
source to the receiver. In order to correctly determine the
frequency content of an acoustic signal, the receiver needs to
compensate for these distortions. We first investigated whether P.
discolor compensates for distortions of the spectral shape of
transmitted sounds during passive listening. Bats were trained to
discriminate lowpass filtered from highpass filtered acoustic
impulses, while hearing a continuous white noise background with a
flat spectral shape. We then assessed their spontaneous
classification of acoustic impulses with varying spectral content
depending on the background’s spectral shape (flat or lowpass
filtered). Lowpass filtered noise background increased the
proportion of highpass classifications of the same filtered
impulses, compared to white noise background. Like humans, the bats
thus compensated for the background’s spectral shape. In an
active-acoustic version of the identical experiment, the bats had
to classify filtered playbacks of their emitted echolocation calls
instead of passively presented impulses. During echolocation, the
classification of the filtered echoes was independent of the
spectral shape of the passively presented background noise.
Likewise, call structure did not change to compensate for the
background’s spectral shape. Hence, auditory processing differs
between passive and active hearing, with echolocation representing
an independent mode with its own rules of auditory spectral
analysis. The second project (chapter 3) was concerned with the
accurate measurement of the time of occurrence of auditory signals,
and as such also distance in echolocation. In addition, the
importance of passive listening compared to echolocation turned out
to be an unexpected factor in this study. To measure the distance
to objects, called ranging, bats measure the time delay between an
outgoing call and its returning echo. Ranging accuracy received
considerable interest in echolocation research for several reasons:
(i) behaviourally, it is of importance for the bat’s ability to
locate objects and navigate its surrounding, (ii) physiologically,
the neuronal implementation of precise measurements of very short
time intervals is a challenge and (iii) the conjectured
echo-acoustic receiver of bats is of interest for signal
processing. Here, I trained the nectarivorous bat Glossophaga
soricina to detect a jittering real target and found a biologically
plausible distance accuracy of 4–7 mm, corresponding to a temporal
accuracy of 20–40 μs. However, presumably all bats did not learn to
use the jittering echo delay as the first and most prominent cue,
but relied on passive acoustic listening first, which could only be
prevented by the playback of masking noise. This shows that even a
non-gleaning bat heavily relies on passive acoustic cues and that
the measuring of short time intervals is difficult. This result
questions other studies reporting a sub-microsecond time jitter
threshold. The third project (chapter 4) linked the perception of
echo-acoustic stimuli to the appropriate behavioural reactions,
namely evasive flight manoeuvres around virtual objects presented
in the flight paths of wild, untrained bats. Echolocating bats are
able to orient in complete darkness only by analysing the echoes of
their emitted calls. They detect, recognize and classify objects
based on the spectro-temporal reflection pattern received at the
two ears. Auditory object analysis, however, is inevitably more
complicated than visual object analysis, because the
one-dimensional acoustic time signal only transmits range
information, i.e., the object’s distance and its longitudinal
extent. All other object dimensions like width and height have to
be inferred from comparative analysis of the signals at both ears
and over time. The purpose of this study was to measure perceived
object dimensions in wild, experimentally naïve bats by
video-recording and analysing the bats’ evasive flight manoeuvres
in response to the presentation of virtual echo-acoustic objects
with independently manipulated acoustic parameters. Flight
manoeuvres were analysed by extracting the flight paths of all
passing bats. As a control to our method, we also recorded the
flight paths of bats in response to a real object. Bats avoided the
real object by flying around it. However, we did not find any
flight path changes in response to the presentation of several
virtual objects. We assume that the missing spatial extent of
virtual echo-acoustic objects, due to playback from only one
loudspeaker, was the main reason for the failure to evoke evasive
flight manoeuvres. This study therefore emphasises for the first
time the importance of the spatial dimension of virtual objects,
which were up to now neglected in virtual object presentations.