Latencies of conditioned vocal responses to hearing test tones in killer whales (Orcinus orca)

Abstract

Introduction: Perceived loudness is challenging to study in non-human animals. However, reaction time to an acoustic stimulus is a useful behavioral proxy for the assessment of perceived loudness. Understanding the effect of sound frequency and level on perceived loudness would improve prediction and modeling of anthropogenic noise impacts on marine mammals.

Methods: In this study, behavioral hearing tests conducted with two killer whales were analyzed to capture conditioned vocal response latency, which is the time between the onset of the acoustic signal and the onset of the response (i.e., reaction time).

Results: The results showed that vocal reaction times decreased with increasing sensation level (i.e., sound pressure level above the baseline hearing threshold), while the effect of frequency on reaction time varied between the subjects. Reaction time as a function of sound duration is described, and equal-latency contours are presented.

Discussion: The data suggest that vocal reaction time decreases with increasing sensation level, therefore supporting the use of reaction time as a proxy for loudness perception in killer whales.

Keywords: equal latency; loudness; marine mammal; reaction time; sensation level.

Figure 1

Spectrogram of a single trial showing the hearing test signal and vocal response. Each trial was recorded at a sample rate of 50 kHz. The onset of the stimulus relative to the start of the recording was always 500 ms. The vocal response latency in this example was 551 ms after the onset of the 100 ms duration signal, and 1,051 ms after the recording had begun. Color indicates relative acoustic intensity with brighter colors representing higher intensities.

Figure 2

Behavioral audiograms for each whale subject. Solid circles and triangles represent behavioral hearing thresholds for 500 ms duration pure tones. All data was from Branstetter et al. (2023) except for thresholds at 5 and 100 kHz for Whale E (Branstetter et al., 2017). Vertical bars represent the variability in hearing thresholds within a frequency due to the effect of signal duration (see Supplementary Tables S1, S2). Signal duration was accounted for by analyzing sensation level instead of sound pressure level as an independent variable.

Figure 3

Median response latency (RT) as a function of frequency for each whale. Error bars represent interquartile ranges.

Figure 4

Median response latency as a function of sensation level for Whale C at each hearing test frequency. Error bars show interquartile range. All median response latency values were binned to the nearest 5 dB SL value (i.e., RT values at a SL of 8 dB were included in the 10 dB SL bin). A generalized linear model showed a significant decrease in response latency with increasing sensation level (*p < 0.05) for all frequencies except 80 kHz. Sample size varied between signal conditions.

Figure 5

Median response latency as a function of sensation level for Whale E at each hearing test frequency. Error bars show interquartile range. All median response latency values were binned to the nearest 5 dB SL value (i.e., RT values with a SL of 8 dB were part of the 10 dB SL bin). A generalized linear model showed a significant decrease in response latency with increasing sensation level (*p < 0.05) at 10, 40, and 80 kHz. Sample size varied between signal conditions.

Figure 6

Examples of the change in response latency as a function of sound pressure level (SPL). Each panel is a single hearing test session where the response latency is plotted for sequential trials. SPL decreased over the course of the hearing test, so SPL decreases from left to right in each panel. Whale C is shown in the top row examples. Whale E is shown in the bottom row examples.

Figure 7

Equal latency contours for Whale C across frequencies. The vertical axis sensation levels are relative to the 20 kHz response latency data indicated by the black arrow. Points along the same line indicate the sensation level at which the response latency is equal to the 20 kHz response latency (resulting from interpolation from the Pieron function described in Equation 1). For example, a 30 dB sensation level signal at 20 kHz would be predicted to have the same response latency as a 36 dB sensation level signal at 10 kHz and a 15 dB sensation level signal at 80 kHz.