Acoustic noise is ubiquitous: it is generated, for example, by wind and moving water, vehicles on the street, people speaking at social gatherings, and diverse species communicating in tropical rain forests. Nevertheless, humans and other animals manage to operate in the presence of competing acoustic signals. One prevailing strategy is to increase the volume of sound production in noisy environments, commonly known as the Lombard effect. Despite over a century of research, how our brain uses noise information to guide our vocal amplitude control has remained elusive. Here, we show that the Lombard effect is a direct response to background noise and features an extremely short latency of about 30 ms. Thus, the Lombard effect can be considered a sensorimotor reflex.
Figure: Echolocating bat tracks a tethered insect in quiet and in the presence of noise bursts. The inset plots rapid increases in sonar call volume in response to noise at different amplitudes.
In this study, we took advantage of the brief sonar vocalizations of big brown bats and mapped out the time course of the Lombard effect on a millisecond time scale. Bats were trained to rest on a platform and track by echolocation tethered insects moving on a pulley system. As bats performed the sonar- tracking task, we presented noise bursts of varying temporal patterns to perturb auditory feedback through a custom ‘bat-computer interface’. Each bat’s vocalizations were recorded by an array of 14 microphones, from which we could precisely quantify call amplitude.
Two key findings emerged from these behavioral experiments. First, the Lombard effect in the bat is extremely fast, 2-3 orders of magnitude faster than current estimates from other animals. Second, the magnitude of the Lombard effect does not merely depend on noise amplitude, as previously believed, but also on noise duration and silence gaps between noise bursts. The high temporal resolution data of this study provided the foundation for the first computational model of the Lombard effect. Surprisingly, the Lombard effect can be explained by a single auditory process, temporal summation, and the entire model contains only three free parameters.
Our study not only demonstrates that the Lombard effect in bats features an extremely short response latency of about 30 ms, but also offers a mechanistic explanation for an observation that has perplexed the scientific community: how could the Lombard effect arise in animals as diverse as fish, frogs, birds, and mammals, considering the wide differences in their hearing systems. The simple answer is that auditory temporal summation operates across taxa.
Reference
Sensorimotor integration on a rapid time scale. Jinhong Luo, Ninad Kothari, Cynthia F. Moss. 2017. Proceedings of the National Academy of Sciences. 114: 6605-6610.
Related articles in the press
Press release from Johns Hopkins University