Communication breakdown: Limits of spectro-temporal resolution for the perception of bat communication calls

Stephen Gareth Hörpel,Uwe Firzlaff,Jonas Reijniers,A Leonie Baier,Herbert Peremans,Lutz Wiegrebe

doi:10.1038/s41598-021-92842-4

Abstract

During vocal communication, the spectro-temporal structure of vocalizations conveys important contextual information. Bats excel in the use of sounds for echolocation by meticulous encoding of signals in the temporal domain. We therefore hypothesized that for social communication as well, bats would excel at detecting minute distortions in the spectro-temporal structure of calls. To test this hypothesis, we systematically introduced spectro-temporal distortion to communication calls of Phyllostomus discolor bats. We broke down each call into windows of the same length and randomized the phase spectrum inside each window. The overall degree of spectro-temporal distortion in communication calls increased with window length. Modelling the bat auditory periphery revealed that cochlear mechanisms allow discrimination of fast spectro-temporal envelopes. We evaluated model predictions with experimental psychophysical and neurophysiological data. We first assessed bats’ performance in discriminating original versions of calls from increasingly distorted versions of the same calls. We further examined cortical responses to determine additional specializations for call discrimination at the cortical level. Psychophysical and cortical responses concurred with model predictions, revealing discrimination thresholds in the range of 8–15 ms randomization-window length. Our data suggest that specialized cortical areas are not necessary to impart psychophysical resilience to temporal distortion in communication calls.

Highlights

During vocal communication, the spectro-temporal structure of vocalizations conveys important contextual information
Quantification of the amount of distortion of the modulation patterns reveals the following: In the aggression calls, the spectrum of amplitude modulation rates changes as a function of phase-randomization window-length (Fig. 2 left)
We have determined the degree of distortion that impedes the recognition of communication calls in the bat Phyllostomus discolor

Summary

Introduction

The spectro-temporal structure of vocalizations conveys important contextual information. We hypothesized that for social communication as well, bats would excel at detecting minute distortions in the spectro-temporal structure of calls. The overall degree of spectro-temporal distortion in communication calls increased with window length. Psychophysical and cortical responses concurred with model predictions, revealing discrimination thresholds in the range of 8–15 ms randomization-window length. Our data suggest that specialized cortical areas are not necessary to impart psychophysical resilience to temporal distortion in communication calls. Studies that investigate the sensory processing of vocal communication signals, including human speech, typically make use of systematic degradations of the vocalizations’. Speech signals had to be analysed with integration time on this scale to ensure efficient decoding These findings were considered to support the notion that a precise analysis of the local acoustic spectrum is non-essential to the speech code[6,8,9]. Global cues such as slow modulation envelopes in the order of 3–8 Hz might be crucial to speech intelligibility in humans and may be the reason for the resilience of speech to distortion of short segments (< 50 ms)[6,9,10]

Methods

Results

Conclusion