Fig 1.
Signature whistle of the dolphin SAY.
(A) Waveform of the whistle and (B) spectrogram of the same whistle displaying the frequency contour that dolphins use for recognition. The whistle contains several harmonics.
Fig 2.
Baseline objects and their associated sounds.
(A) “Rope” was a linear upsweep frequency contour. (B) “bottle” was a repeating linear upsweep, while (C) “ball” was a single cycle of a sinusoid. All baseline sounds were 500 ms in duration with equal bandwidths between 8 kHz and 12 kHz, with a center frequency of 10 kHz.
Fig 3.
Examples of transformed “ball” test sounds.
(A) ball has been transposed ½ octave down (re: baseline stimulus) in frequency, (B) shifted ½ octave up in frequency, and (C) the duration has been halved to 250 ms.
Table 1.
Test sounds used in the dolphin experiment.
There were three sound types (rope, bottle and ball). Each sound type had three sound dimensions (amplitude, frequency and duration). And each sound dimension had five different dimension levels that were tested, for a total of 45 different test sounds. All underwater dB levels had a reference level of 1 μPa. Frequency values reflect the center frequency of each sound.
Table 2.
Test sounds used in the human experiment.
Each sound type had three sound dimensions (amplitude, frequency and duration). And each sound dimension had five different dimension levels that were tested, for a total of 45 different test sounds. All in-air dB levels had a reference level of 20 μPa.
Fig 4.
(A) Above view and (B) underwater view. During each trial, the dolphin stationed on an underwater biteplate at a 2 m radial distance from a nylon rope, a water filled aluminum bottle and an air filled steel sphere. Each of the three objects was suspended by monofilament at the same depth as the bite plate. The locations of the objects never changed.
Fig 5.
Proportion correct as a function of (A) SPL (amplitude), (B) center frequency, and (C) sound duration. Each gray data point represents proportion correct for ten trials and black data points represent averages.
Table 3.
Errors made by the dolphin participant on each transformation type.
The sound played is displayed in the left column of each matrix, and the dolphin’s choice is displayed along the top row. The sum column indicates the total incorrect trials for each sound type
Table 4.
Immediacy of recognition by a dolphin of novel test sounds.
An “X” indicates a correct choice the very first time the dolphin was presented with the stimulus, while an empty space indicates an incorrect choice for the first time.
Fig 6.
Proportion correct as a function of (A) SPL, (B) center frequency, and (C) sound duration. Each gray data point represents proportion correct for ten trials and black data points represent averages. Note that the y-axis scale is different for Fig 5 and Fig 6.
Fig 7.
Comparison of errors (confusions between sound types) made by the human participants and the dolphin for each sound transformation.
Table 5.
Errors made by human participants (N = 30) on each transformation type.
The sound played is displayed in the left column of each matrix, and the participant’s choice is displayed along the top row. The sum column indicates the total incorrect trials for each sound type.
Table 6.
Immediacy of recognition with novel test sounds for the human participants and the dolphin subject.
Bold text indicates the dolphin was correct on her first trial, while regular text indicate the dolphin was incorrect on her first trial. The percent of human participants (N = 30) that achieved a correct answer on the first trial for each sound type and transformation is shown.