Previous studies suggested that dolphins perceive echo spectral features on coarse (macrospectrum) and fine (microspectrum) scales. This study was based on a finding that these auditory percepts are, to some degree, dependent on the dolphin's ∼250-μs auditory temporal window (i.e., “critical interval”). Here, two dolphins were trained to respond on passively detecting a simulated “target” echo complex [a pair of echo “highlights” with a characteristic 120-μs inter-highlight interval (IHI)]. This target had unique micro- and macrospectral features and was presented among “distractor” echoes with IHIs from 50 to 500 μs (i.e., microspectra) and various highlight durations (i.e., macrospectra). Following acquisition of this discrimination task, probe echo complexes with the macrospectrum of the target but IHIs matching the distractors were infrequently presented. Both dolphins initially responded more often to probes with IHIs of 80–200 μs. Response strategies diverged with increasing probe presentations; one dolphin responded to a progressively narrower range of probe IHIs while the second increased response rates for probes with IHIs > 250 μs. These results support previous conclusions that perception of macrospectra for complex echoes is nonconstant as the IHI decreases below ∼100 μs, but results approaching and exceeding 250 μs—the temporal window upper boundary—were more ambiguous.

Echolocation clicks produced by dolphins interact with physical objects to generate complex echoes that can be used to identify, classify, and discriminate objects such as fish and other underwater targets of interest (Au, 1993; Helweg et al., 2003; Au et al., 2009). Each reflective surface of a target object creates an echo “highlight” resembling the incident biosonar signal. The combination of these highlights, which may have slightly different time delays depending on the spatial arrangement of the target's reflective surfaces, comprises the resultant complex echo (Au and Hammer, 1980).

How the characteristics of complex echoes affect object perception by odontocetes has been of interest in terms of sensory ecology and psychology (Vel'min and Dubrovsky, 1976b; Hammer and Au, 1980; Moore et al., 1984; Simmons et al., 2014) and bio-inspired sonar applications, such as automatic target recognition (ATR) technology (Pailhas et al., 2012). Due to the refractory nature of primary auditory neurons, the odontocete ear is not capable of resolving all rapidly occurring highlights following the onset of a complex echo (Supin and Popov, 1995b,a; Popov et al., 2001). Consequently, explanations of dolphins' abilities to detect the small-scale spatial differences of reflectors within target objects have instead focused on interactions among echo highlights separated by some time delay (Au and Hammer, 1980; Moore et al., 1984; Au et al., 1988; Au and Pawloski, 1992; Branstetter et al., 2020). These correlated echo highlights produce spectral interference patterns that resemble a repetition of peaks and notches in the spectrum of the complex echo according to Δf = 1/Δt, where Δf is the spacing of the repetitive peaks and notches in the overall echo spectrum, and Δt is the time separation of the two highlights, often called the “inter-highlight interval” (IHI; Au, 1993; Ming et al., 2021). In dolphins, the ability to discern the IHI in complex echoes has also been likened to the detection of a time-separation pitch (TSP), also called repetition pitch (Au and Hammer, 1980; Au and Pawloski, 1989), where, for humans, the percept of the sound is “colored” by a pitch related to the inverse of Δt (Thurlow and Small, 1955; Bilsen and Ritsma, 1969; Yost and Hill, 1978). In a detailed model of bat biosonar, the spectrogram correlation and transformation (SCAT) model, the transformation of Δf into Δt, is accomplished through a delay line system, allowing for a reconstruction of the original highlight time separation and, thus, the fine-scale features of the target (Saillant et al., 1993). The recent application of the SCAT model to dolphin biosonar signals further reinforces the potential importance of spectral interference patterns for reconstructing highlight separations in target classification (Ming et al., 2021).

A series of behavioral experiments with the bottlenose dolphin [(Tursiops truncatus; Dubrovsky et al., 1992), and later reported by Zorikov and Dubrovsky (2003)] examined the perception of spectral interference patterns (termed the microspectrum by the authors) in simulated two-highlight echo complexes as they related to other acoustic features, including the coarse spectral envelope (termed the macrospectrum). The experiments demonstrated that while the macrospectrum was the perceptually dominant feature, dolphins also attended to microspectral cues in classifying stimuli (Vel'min and Dubrovsky, 1976b; Dubrovsky et al., 1992). This result was supported recently by a similar study with dolphins (Accomando et al., 2020). For microspectral cues to be preserved, echo highlights should fall within an auditory temporal window where successive echo highlights are “integrated” in a manner analogous to a Fourier transform containing both highlights. In bottlenose dolphins, the auditory temporal window duration (“critical interval”) during which such a process might take place has been identified as approximately 250–300 μs. This has been demonstrated using behavioral (Vel'min and Dubrovsky, 1976b,a; Moore et al., 1984; Au et al., 1988; Dubrovsky, 1990; Dubrovsky et al., 1992; Branstetter et al., 2020) and electrophysiological methods (Supin and Popov, 1995b; Popov and Supin, 1997; Popov and Supin, 1998; Popov et al., 2001; Finneran et al., 2020b).

Dubrovsky et al. (1992) conducted an additional experiment regarding microspectrum perception in which the IHI was manipulated across values spanning the upper limit of the temporal window. Dolphins were first trained to respond to a positive signal—a simulated two-highlight echo—with unique durations of the individual highlights and a 120-μs IHI. The negative signal (i.e., dolphins were trained to withhold response) was comprised of two unique individual highlights that are different from the positive signal but also separated by 120 μs. Probe signals were then introduced with variable IHI from 50 to 500 μs. These probes were otherwise identical to the positive signal, and the dolphin was rewarded for responding to their presentation. The authors calculated a ≥75% response rate for probes between ∼100 and 200 μs and concluded that macrospectrum feature perception was dependent on the IHI; at IHIs of less than ∼100 μs or greater than ∼200 μs, the lower response rates indicated that the perception of a macrospectrum equivalent to the positive signal was altered. Distortions in the macrospectrum below the ∼100 μs boundary likely occurred due to an increase in periodicity of the microspectrum structure sufficient to render it perceived as macrospectrum, whereas IHIs greater than ∼200 μs resulted in perception of highlights as distinct elements, which is in accordance with prior findings and roughly aligning with the duration of the dolphin's 250-μs temporal window. Similar results were found by Shriram and Simmons (2019) with the big brown bat (Eptesicus fuscus). In a two-alternative forced-choice task, bats found negative (i.e., unreinforced) targets with IHIs of 36–300 μs to be perceptually similar to a positive (i.e., reinforced) 100-μs IHI standard. This upper limit corresponds roughly to the 350-μs auditory temporal window in bats (Simmons et al., 1988; Sanderson and Simmons, 2000; Shriram and Simmons, 2019). Together, these findings suggest that complex target echoes, where the two highlights arrive within some upper and lower auditory temporal window boundaries, will be classified by the listener as having similar macrospectra.

The current experiment explores the interaction of micro- and macrospectrum features of complex echoes through an elaboration on the Dubrovsky et al. (1992) experiment described above. Two dolphins were trained to passively listen to and discriminate simulated, two-highlight echoes with various combinations of macro- and microspectral features. The goals were to further examine the nature of macro- and microspectrum perception across IHIs spanning the duration of the bottlenose dolphin temporal window. Specifically, the current study included a larger set of “distractor” stimuli with various macro- and microspectra with the goal of increasing the confidence that dolphins' responses to probe stimuli arose from perceptual similarity with the trained positive stimulus.

Two bottlenose dolphins, LRK (female, 16 years of age) and APO (male, 5 years of age), participated in the study. The upper-frequency limit of hearing—determined from auditory evoked potential measurements—was approximately 140 kHz for LRK and 80 kHz for APO (Strahan et al., 2020). For this species, individuals with the full bandwidth of hearing have an upper-frequency cutoff of >120 kHz. Thus, while LRK had a full range of hearing, APO was classified as having high-frequency hearing loss. Data were collected between December 2020 and March 2021 and followed a protocol approved by the Institutional Animal Care and Use Committee (IACUC) of the Naval Information Warfare Center (NIWC) Pacific and the Navy Bureau of Medicine and Surgery. It followed all applicable U.S. Department of Defense guidelines for the care and use of animals.

Sessions were conducted in a 9 m × 9 m floating netted enclosure located at the U.S. Navy Marine Mammal Program in San Diego Bay, CA. Dolphins stationed on a submerged biteplate supported by an aluminum frame, which was attached to a deck-spanner in the center of the enclosure. An underwater sound projector was attached to the frame and positioned ∼1 m in front of the biteplate (see Fig. 1 in Finneran et al., 2020a, for a schematic of the setup). A response paddle was located approximately 3 m directly in front of the biteplate at the same depth (∼1 m). To press the paddle, the dolphin was required to swim around or underneath the aluminum frame. A second underwater sound projector (DAEX25W-8 waterproof exciter, Dayton Audio, Springboro, OH) was also suspended from the aluminum frame to play conditioned “correct” or “incorrect” sounds, which provided auditory feedback to the dolphins regarding individual trial performance.

The dolphins were trained to passively listen to simulated two-highlight echo complexes. The two-highlight complex is referred to as an “echo” hereafter. One class of echoes were distractors with various macrospectra arising from the duration of the square (DC, direct current) pulses used to drive the sound projector (see below) and various microspectra arising from IHIs between 50 and 500 μs (see Table I). “Target” echoes, which dolphins were trained to discriminate, had a set macrospectrum arising from the fixed DC pulse durations of 9 and 14 μs and microspectrum arising from a fixed 120-μs IHI. The experimental task was a “go/no-go” paradigm. Each trial began with the presentation of blocks of distractor echoes as the dolphin was instructed by the trainer to swim to the biteplate. Once the dolphin was on the biteplate, a trial was initiated with the presentation of 2–5 1-s “blocks” of 20 distractor echoes, where each block was a repeated distractor echo and each successive block used a different distractor echo (see below). Next, either a 1-s block of 20 target echoes (“target trial”) or a final block of 20 distractor echoes (“control trial”) was presented. The dolphin was required to touch the response paddle (“go”) in response to target echoes and to remain stationary (“no-go”) in the absence of target echoes. Following acquisition of this discrimination task, a small proportion of trials each day (“probe trials”) featured a 1-s block of 20 “probe” echoes in lieu of target echoes or the final block of distractor echoes. Probe echoes had the same macrospectrum as target echoes but IHIs varied from 50 to 500 μs (i.e., varied microspectrum similar to distractors; see Table I). The primary data of interest were the dolphins' responses to the probe trials—how would probe microspectrum affect the dolphin's perception of a probe echo as similar or dissimilar to the target echoes?

TABLE I.

The parameters used to generate distractor, target, and probe echoes. All of the echoes were generated from pairs of DC pulses separated by an IHI (i.e., one echo = one pair of pulses; see Fig. 2). For distractor echoes, two pulse durations were randomly paired with a random IHI. For target and probe echoes, either the 9- or 14-μs pulse durations could be presented first. For a description of the sequence of echoes for each trial, see Fig. 3.

Echo typeDC pulse durations (μs)IHI (μs)
Distractor 6, 12, 17, 20, 23, 26 50–500 
Target 9, 14 120 
Probe 9, 14 50, 70, 80, 90, 105, 140, 160, 180, 200, 250, 300, 400, 500 
Echo typeDC pulse durations (μs)IHI (μs)
Distractor 6, 12, 17, 20, 23, 26 50–500 
Target 9, 14 120 
Probe 9, 14 50, 70, 80, 90, 105, 140, 160, 180, 200, 250, 300, 400, 500 

The electrical waveforms used to generate simulated echoes were paired, DC pulses with durations ranging from 6 to 26 μs (see Fig. 1 and Table I) and separated by IHIs of 50–500 μs (see Fig. 2 and Table I). Stimuli were generated using custom LabVIEW-based software (National Instruments Corporation, Austin, TX) and converted to analog (1 MHz) with a National Instruments PCI-6251 multifunction data acquisition (DAQ) card. The pulses were filtered (200 Hz–200 kHz bandpass, 3C filter module, Krohn-Hite Corporation, Brockton, MA), routed through an SR560 preamplifier (Stanford Research Systems, Sunnyvale, CA), and amplified with a 7602M power amplifier (Krohn-Hite Corporation). These produced broadband acoustic transients when transmitted from the underwater projector (ITC 5446 directional piezoelectric transducer, International Transducer Corp., Santa Barbara, CA).

FIG. 1.

(Color online) Example waveforms (left) and spectra (right) of individual echo highlights demonstrating the dependence of larger scale spectral features (macrospectrum) on the duration of the DC pulse used to generate the highlight (listed on each waveform, left). As the duration of the DC pulse is increased, the spectrum presents more closely spaced notches. The 9- and 14-μs pulses (blue and red traces, respectively) were paired to create the target echo while the other durations (black traces) were used in constructing distractor echoes (see Fig. 2 for waveforms and spectra of complex echoes).

FIG. 1.

(Color online) Example waveforms (left) and spectra (right) of individual echo highlights demonstrating the dependence of larger scale spectral features (macrospectrum) on the duration of the DC pulse used to generate the highlight (listed on each waveform, left). As the duration of the DC pulse is increased, the spectrum presents more closely spaced notches. The 9- and 14-μs pulses (blue and red traces, respectively) were paired to create the target echo while the other durations (black traces) were used in constructing distractor echoes (see Fig. 2 for waveforms and spectra of complex echoes).

Close modal
FIG. 2.

(Color online) Example waveforms (left) and spectra (right) of 9- and 14-μs two-highlight pairs with various IHIs. The IHI is listed on each waveform, including the 120-μs target IHI and example probes (see Table I for all of the probe conditions). As the IHI increases, the overall spectral envelope (macrospectrum) characteristic to the 9- and 14-μs pair is constant, but the spacing of small-scale notches (microspectrum) decreases. For reference, the spectra from the individual 9- and 14-μs pulses from Fig. 1 are overlaid on the overall spectrum for the 120-μs IHI target echo (blue dashed line for the 9-μs spectrum and red dotted line for the 14-μs spectrum).

FIG. 2.

(Color online) Example waveforms (left) and spectra (right) of 9- and 14-μs two-highlight pairs with various IHIs. The IHI is listed on each waveform, including the 120-μs target IHI and example probes (see Table I for all of the probe conditions). As the IHI increases, the overall spectral envelope (macrospectrum) characteristic to the 9- and 14-μs pair is constant, but the spacing of small-scale notches (microspectrum) decreases. For reference, the spectra from the individual 9- and 14-μs pulses from Fig. 1 are overlaid on the overall spectrum for the 120-μs IHI target echo (blue dashed line for the 9-μs spectrum and red dotted line for the 14-μs spectrum).

Close modal

The target echo was always a 9- and 14-μs pulse pair (in either order as the order did not affect the overall target spectra) separated by a 120-μs IHI (Fig. 2). Probe echoes used the same 9- and 14-μs pulse pairs as the target but featured 13 discrete IHI values between 50 and 500 μs (see Table I). Each distractor echo was generated using two randomly selected (with replacement) pulse durations from Table I with the IHI randomly selected between 50 and 500 μs (uniform distribution). However, due to an error in the selection code, the probability of occurrence for the 6- and 26-μs distractor echo pulse durations was half of that for the other four pulse durations. All of the echoes were presented in blocks of 20 with an inter-echo interval of 50 ms (resulting in a 1-s block; Fig. 3) with blocks of distractor echoes preceding a single block of either target or probe echoes. Within a 1-s block, the pulse durations and IHI were fixed. On control trials, only blocks of distractor echoes were played. Echoes were presented at a mean received sound exposure level (SEL) of 108 dB re 1 μPa2 s with echo SELs roved ±3 dB to eliminate the potential for any perceptual loudness cues between conditions.

FIG. 3.

(Color online) An example of the target trial sequence. Four distinct distractor echoes (shades of blue; i.e., the top left waveform is one of the four two-highlight echoes) are presented sequentially, 20 times each (a “block,” black outlines), and roved ± 3 dB to remove amplitude cues. The example distractor echo (top left) was generated using a pair of 6- and 23-μs DC pulses (macrospectrum), separated by a 470-μs IHI (microspectrum). In a control trial, only distractor echo blocks would be played. The target echo (right), a 9- and 14-μs DC pulse pair (in either order) separated by 120 μs (see Fig. 2), remained the same and was roved ± 3 dB to remove amplitude cues. The dolphins were trained to withhold response to the distractor echoes and respond if the target echo (red, right waveform above) was heard. For probe trials, target echoes were replaced with paired 9- and 14-μs DC durations having IHIs different from the 120-μs target IHI (see Table I) to examine the effect of IHI on micro- and macrostructure perception.

FIG. 3.

(Color online) An example of the target trial sequence. Four distinct distractor echoes (shades of blue; i.e., the top left waveform is one of the four two-highlight echoes) are presented sequentially, 20 times each (a “block,” black outlines), and roved ± 3 dB to remove amplitude cues. The example distractor echo (top left) was generated using a pair of 6- and 23-μs DC pulses (macrospectrum), separated by a 470-μs IHI (microspectrum). In a control trial, only distractor echo blocks would be played. The target echo (right), a 9- and 14-μs DC pulse pair (in either order) separated by 120 μs (see Fig. 2), remained the same and was roved ± 3 dB to remove amplitude cues. The dolphins were trained to withhold response to the distractor echoes and respond if the target echo (red, right waveform above) was heard. For probe trials, target echoes were replaced with paired 9- and 14-μs DC durations having IHIs different from the 120-μs target IHI (see Table I) to examine the effect of IHI on micro- and macrostructure perception.

Close modal

Echoes were presented with continuous, spectrally white masking noise (mean pressure spectral density level = 70 dB re 1 μPa2 s/Hz) from 20 to 200 kHz. A 3-s sample of noise was continuously generated with a DAQ card (National Instruments USB-6251, 1-MHz update rate, Austin, TX), lowpass filtered (300 kHz, –3 dB) using the SR560 preamplifier, amplified using the 7602M power amplifier, and delivered using the same ITC 5446 projector that was used for echoes. The noise was digitally compensated to account for the transmitting response and any multipath interference, creating a nominally flat noise floor to provide a consistent signal-to-noise ratio (see Fig. 4) over the ambient noise in San Diego Bay (normally dominated by snapping shrimp, vessel traffic, and other dolphins) and within the hearing bandwidth of the subject dolphins. The noise was played for the duration of each session. Echoes and background noise were calibrated using a TC4013 hydrophone (Reason Inc., Slangerup, Denmark) positioned at a location corresponding to the dolphin's sound reception region near the midpoint of the posterior mandible with the dolphin absent.

FIG. 4.

An example spectrum of the received masking noise that was projected over ambient noise in San Diego Bay during testing. Experimentally controlled Gaussian white noise from 20 to 200 kHz was continually played back at a mean level of 70 dB re 1 μPa2/Hz for the duration of each session. The increase in noise levels near 160 kHz is due to the resonance frequency of the ITC 5446 projector; however, the majority of this spectral region is above the upper-frequency hearing limits of the dolphins (see Sec. II).

FIG. 4.

An example spectrum of the received masking noise that was projected over ambient noise in San Diego Bay during testing. Experimentally controlled Gaussian white noise from 20 to 200 kHz was continually played back at a mean level of 70 dB re 1 μPa2/Hz for the duration of each session. The increase in noise levels near 160 kHz is due to the resonance frequency of the ITC 5446 projector; however, the majority of this spectral region is above the upper-frequency hearing limits of the dolphins (see Sec. II).

Close modal

1. Training

The dolphins were first trained to perform the paddle-press in response to the target echo. Here, dolphins were directed to station on the underwater biteplate while blocks of distractor echoes were continually presented at very low levels (43 dB re 1 μPa2 s). For target trials, a block of target echoes was then presented at 108 dB re 1 μPa2 s and paired with a visual cue from the trainer for the paddle-press response. Control trials consisted of low-level distractor echoes only. Probe echoes were not presented during training. Typical training sessions consisted of 60 trials. Correct responses (paddle-press in response to target echoes, remaining on the bite plate in the absence of target echoes) were equally reinforced with a “correct” sound and a subsequent fish reward. Incorrect responses were not reinforced and an “incorrect” sound was played back. Over successive sessions and within each session, distractor echo levels were gradually increased until they equaled the target SEL and  ≥90% correct performance was achieved. Once the behavior had been obtained (after a total of 1478 and 1312 trials for LRK and APO, respectively), target and control trials were presented with a 50:50 proportion. The percentage of reinforced correct trials was then reduced in 2% increments from 100% to 90%, which was maintained in all of the later experimental sessions. For unreinforced trials, no acoustic feedback was given (i.e., no correct or incorrect sounds), and no fish reward was given—the dolphin returned to the trainer and was resent to the biteplate for the next trial. Each dolphin completed at least two sessions with ≥90% performance for each 2% reinforcement reduction step. This was done to increase the resiliency of the dolphins to the lack of reinforcement on half of the probe trials in later sessions (see below) and reduce potential associative learning effects due to lack of reinforcement for some probes (as had been previously observed by Accomando et al., 2020).

Once the dolphins achieved correct performance ≥90% for three consecutive sessions in which 90% of the trials were reinforced, data collection with the presentation of probe trials began. During initial data collection, an amplitude cue was discovered that allowed the dolphins to easily discriminate the target echo. On discovery of the amplitude cue, the training process was repeated as described above to properly condition the desired behavior. A time gap of approximately 5 weeks for APO and 7 week for LRK was allowed before initial data collection and retraining, and a total of 349 trials for LRK and 818 trials for APO were required for retraining after removal of the unintended cue. As the dolphins heard each probe condition ten times before the correction of the error, they could not be considered truly naive to the probes during the subsequent experiments described here.

2. Experiment

All probe echo IHIs were tested over the course of two consecutive daily sessions, once per week. Over the course of the study, “maintenance” sessions were also run, consisting of 50–70 trials without probes (50% target, 50% control). One or two maintenance sessions were conducted between sets of experimental sessions, and 90% of correct response trials were reinforced in all of the sessions. Typical sessions lasted between 20 and 40 min. Experimental sessions included 6 or 7 probe trials out of 70–80 total trials, respectively, with 1 probe randomly replacing a target or distractor trial in each block of 10 trials. The first ten trials in each session were considered a “warmup” during which no probe trials were presented. To proceed with the session, dolphins were required to respond correctly to at least eight out of ten warmup trials. If a dolphin did not meet this criterion, the session was continued as a maintenance session, and probes were delayed until a later session. Within each session and each week, probe trial echoes featured a unique IHI. Each probe trial condition was randomly reinforced half of the time over all of the sessions (alternated presentation-to-presentation) to minimize reinforcement-based bias in responses. This weekly procedure was conducted ten times such that ten responses for each probe IHI (Table I) were collected per dolphin.

Both dolphins participated in a total of 20 experimental sessions with LRK and APO participating in 16 and 19 maintenance sessions, respectively. Across experimental and maintenance sessions, combined error rates on target and control trials were low for both APO (1.3%) and LRK (0.04%). Out of 2698 target and control trials, APO made 34 errors (13 false alarms and 21 misses). Out of 2350 target and control trials, LRK made 9 errors (3 false alarms and 6 misses). The low error rates indicated extremely reliable target discrimination from the background echoes. There was no apparent relationship between specific distractor echoes and presence of errors.

Both dolphins responded identically on the first presentation of each probe condition except for the 250-μs IHI (Tables II and III and Fig. 5). Both responded to probes with the IHI between 80 and 200 μs (also 250 μs for LRK) and withheld responses above and below this range (Tables II and III and Fig. 5). APO's response pattern remained relatively stable with increasing probe presentations, but LRK progressively responded to probe echoes with larger IHI. About halfway through the experiment (presentation 6), LRK began responding to nearly all of the probe echoes with IHI ≥ 105 μs with only one exception (see Table III and Fig. 5 left panels). In contrast, APO's responses to probe echoes with IHI > 140 μs decreased with repeated presentation (see Table II and Fig. 5 left panels). Throughout the study, both dolphins consistently responded to 90% or 100% of the probes when the probe IHI was within 20 μs of the target IHI. The two subjects responded similarly over the course of the experiment for probe echoes with the IHI shorter than the 120-μs target echo, as illustrated by the data to the left of the vertical line in Fig. 5.

TABLE II.

The results of probe trials from subject APO. Each row denotes a set of probes (presentation number). The shaded cells with “N” represent no response, and “Y” represents a response.

Presentation numberProbe echo IHI (μs)
50708090105140160180200250300400500
10 
Presentation numberProbe echo IHI (μs)
50708090105140160180200250300400500
10 
TABLE III.

The results of probe trials from subject LRK. Each row denotes a set of probes (presentation number). The shaded cells with “N” represent no response, and “Y” represents a response.

Presentation numberProbe echo IHI (μs)
50708090105140160180200250300400500
10 
Presentation numberProbe echo IHI (μs)
50708090105140160180200250300400500
10 
FIG. 5.

(Color online) The results from LRK (top) and APO (bottom) are shown separately. The number of responses by presentation (n = 10) for all probe IHIs in μs (n = 13; left) is shown. See also Tables II and III. A vertical line is positioned at the target IHI = 120 μs. The proportion of responses to probe echoes (9- and 14-μs DC pulse durations) with various IHIs (right) is depicted. A subset of the results from the current study: presentation 1 (initial probe responses, solid line) and ten presentations (total probe response proportion, dashed line) are compared to those reported by Dubrovsky et al. (1992) (dotted line).

FIG. 5.

(Color online) The results from LRK (top) and APO (bottom) are shown separately. The number of responses by presentation (n = 10) for all probe IHIs in μs (n = 13; left) is shown. See also Tables II and III. A vertical line is positioned at the target IHI = 120 μs. The proportion of responses to probe echoes (9- and 14-μs DC pulse durations) with various IHIs (right) is depicted. A subset of the results from the current study: presentation 1 (initial probe responses, solid line) and ten presentations (total probe response proportion, dashed line) are compared to those reported by Dubrovsky et al. (1992) (dotted line).

Close modal

Both dolphins' response behaviors on the target and control trials were consistently near-perfect (i.e., 100% and 0%, respectively), thus, enabling probe trials to be used as an indicator of whether a particular probe echo was interpreted as being similar to the 120-μs IHI target echo. Since the target echo had distinct macro- and microspectral features while distractor echoes had distinct macrospectral features and microspectral features that overlapped with the target, several possible interpretations of the dolphins' responses to the probe trials emerge. The first, as proposed by Dubrovsky et al. (1992), is that the dolphins primarily based their responses on the macrospectrum when categorizing a probe echo. According to this interpretation, in the IHI range of roughly 100–200 μs, the macrostructure is preserved independent of manipulations in the IHI. The lack of response to probe echoes at 90 μs IHI or less for both dolphins throughout the experiment suggests that the microspectrum at lower IHIs becomes sufficiently large to distort the macrospectrum of the two-highlight echo. Conversely, near the upper limit of the ∼250-μs temporal window, the two highlights do not sufficiently interact and their macrospectra are perceived independently.

An alternative explanation to that proposed by Dubrovsky et al. (1992) is that the probe response functions might instead represent a generalization gradient to the macro- and microspectral qualities of the target. Specifically, the IHIs above and below this range—where the dolphins did not respond—might not necessarily represent a lack of constancy of only the combined 9- and 14-μs macrostructure (as suggested by Dubrovsky et al., 1992) but could plausibly suggest that the specific combination of target macrospectrum and microspectrum were necessary to classify the first probe presentation (see Fig. 5 right panels). Indeed, with continued probe presentation in the absence of consistent reinforcement, the range of probe IHIs to which APO responded progressively narrowed. This result is consistent with increasingly strict criteria for macro- and microstructures similarity relative to the target.

There is evidence to suggest that such a gradient based on the macro- and microstructures of the echo may not be the most parsimonious explanation of the current results. Shriram and Simmons (2019) concluded that the 36–300 μs range over which bats found IHIs to be similar to those of the 100-μs target was notably larger than the measured IHI discrimination abilities of Eptesicus—on the order of 25–30 μs (Simmons et al., 1990; Mogdans et al., 1993). Thus, the observed pattern of responses by bats may reflect IHI categorization based on a “global, rippled character of the spectrum” (Shriram and Simmons, 2019). For dolphins, the discrimination ability while echolocating is on the order of 0.5–1 μs when detecting IHI changes for a two-highlight target echo with 100 μs between highlights (Branstetter et al., 2020).1 The contrast of this acute discrimination capability with the broad distribution of responses for IHIs between 100 and 200 μs from the dolphins in the current study appears in line with the conclusions of Shriram and Simmons (2019).

The range of distractor highlights up to 26 μs in duration (see Fig. 1) might have influenced the sharp decrease in the dolphins' responses to probes with IHIs < 100 μs relative to higher IHIs. The longer-duration DC pulses used to generate highlights in this study resulted in acoustic transients at the start and end of the DC pulse (see longer DC durations in Fig. 1). Spectral notches were, therefore, present at intervals of the reciprocal of 26 μs (Δf = 38 kHz) in this longest duration DC pulse. Although this is almost a one-octave change in spectral notch spacing relative to the shortest IHI of 50 μs (Δf = 20 kHz) and a nearly a two-octave change from the 100-μs IHI region where probe responses began to decline (Δf = 10 kHz), a decrease in probe responses due to spectral similarity to the distractor echoes cannot be completely ruled out.

If the consistent steep decline in the dolphins' responses at the lower probe IHIs is, however, reflective of a distortion of the macrospectrum (as proposed by Dubrovsky et al., 1992) rather than similarity to the longest duration distractors, this result demonstrates a change in perception of complex echoes with short IHIs relative to longer IHIs. This has not been discussed to the same degree as the changes in perception that occur as IHIs approach the upper limit of the temporal window of ∼250 μs. Considering IHIs less than this duration, work has shown that dolphins (1) discriminate multi-highlight echoes based on spectral cues (Au and Hammer, 1980; Au and Pawloski, 1989; Au and Pawloski, 1992; Branstetter et al., 2020), (2) do not display distinct neural (population) responses to highlights subsequent to the first in a complex echo (Supin and Popov, 1995b; Popov et al., 2001), and (3) have detection thresholds that are approximately 3 dB lower than those for IHIs exceeding the temporal window (Au et al., 1988). The results of the current study along with those of Dubrovsky et al. (1992) suggest that the experimental paradigms used by previous studies might not have fully captured a transition in perception that takes place as IHIs are reduced below 100 μs. If these short IHIs result in microspectrum notch spacing that is sufficiently large to take on a macrospectrum quality, perhaps the perception of multiple highlight spacing is degraded and multiple reflectors are not perceived as physically separate. This concept is speculative based on the limited data from the current study and that of Dubrovsky et al. (1992) but is clearly of further interest.

The 250-μs boundary of the dolphins' temporal window may mark the end of the time period associated with microspectrum peaks and notches arising from the echo IHI and the beginning of the period where auditory neurons directly code the timing of subsequent highlights. The fact that LRK responded to nearly all of the probes above 250 μs with repeated presentation indicates that the macrostructure characteristics of the echo pair might still be preserved at these longer intervals and in the likely absence of microspectral cues. However, since APO responded to three out of ten 250-μs IHI probes and did not respond to any probes with IHIs > 250 μs, individual differences in perception and/or strategy must be noted—APO may not have relied on macrospectral cues to the same extent as LRK did, or they might have adopted opposite strategies after repeated probe presentations.

Although the goal of this study was to determine how dolphins resolve complex echoes and provide data comparable to those of Dubrovsky et al. (1992), passive detection tasks may not provide a comprehensive assessment of biosonar capability and processing. An obvious extension of this study would be to conduct a similar experiment but with simulated echoes (targets, distractors, and probes) produced by a phantom echo generator (e.g., Aubauer and Au, 1998; Aubauer et al., 2000; Finneran et al., 2020b) during an active biosonar task. Further, a systematic modification of the target IHI and/or the distractor highlight durations—either in a biosonar or passive listening task—would help to determine if the observed probe responses would be similar when tested with a different experimental design.

The current findings further demonstrate that the interplay between the fine-scale acoustic characteristics of complex echoes and basic properties of the dolphin auditory system (i.e., the temporal window) influences perception at a higher level. This ∼250-μs auditory temporal window continues to appear as a central feature in a diverse group of dolphin hearing and echolocation studies (Vel'min and Dubrovsky, 1976b,a; Moore et al., 1984; Au et al., 1988; Dubrovsky, 1990; Dubrovsky et al., 1992; Supin and Popov, 1995a; Popov and Supin, 1997, 1998; Popov et al., 2001; Branstetter et al., 2020; Finneran et al., 2020b). The apparent influence of the temporal window in this study's complex behavioral task contributes another piece to the puzzle of dolphin echolocation as a whole: from the basic encoding of acoustic features arising from a target's physical parts to discerning and categorizing its shape.

The authors thank the animal training staff without whom this research could not be successful, K. Donohoe and E. McGarvey for assistance running maintenance sessions, and C. Reichmuth for experimental design advice. We acknowledge funding from the Office of Naval Research Code 32 (Mine Countermeasures, Acoustics Phenomenology and Modeling Group), and contributions from two anonymous reviewers, which improved this manuscript. This is contribution number 338 of the National Marine Mammal Foundation.

1

In our recent experiments using a passive IHI discrimination task and 120-μs IHI target, this ability is only slightly worse, near 2 μs.

1.
Accomando
,
A. W.
,
Mulsow
,
J.
,
Houser
,
D. S.
, and
Finneran
,
J. J.
(
2020
). “
Classification of biosonar target echoes based on coarse and fine spectral features in the bottlenose dolphin (Tursiops truncatus)
,”
J. Acoust. Soc. Am.
148
,
1642
1646
.
2.
Au
,
W. W. L.
(
1993
).
The Sonar of Dolphins
(
Springer
,
New York
).
3.
Au
,
W. W. L.
,
Branstetter
,
B. K.
,
Benoit-Bird
,
K. J.
, and
Kastelein
,
R. A.
(
2009
). “
Acoustic basis for fish prey discrimination by echolocating dolphins and porpoises
,”
J. Acoust. Soc. Am.
126
,
460
467
.
4.
Au
,
W. W. L.
, and
Hammer
,
C. E.
(
1980
). “
Target recognition via echolocation by Tursiops truncatus
,” in
Animal Sonar Systems
, edited by
R. G.
Busnel
and
J. F.
Fish
(
Plenum
,
New York
), pp.
855
858
.
5.
Au
,
W. W. L.
,
Moore
,
P. W. B.
, and
Pawloski
,
D. A.
(
1988
). “
Detection of complex echoes in noise by an echolocating dolphin
,”
J. Acoust. Soc. Am.
83
,
662
668
.
6.
Au
,
W. W. L.
, and
Pawloski
,
J. L.
(
1989
). “
Detection of noise with rippled spectra by the Atlantic bottlenose dolphin
,”
J. Acoust. Soc. Am.
86
,
591
596
.
7.
Au
,
W. W. L.
, and
Pawloski
,
D. A.
(
1992
). “
Cylinder wall thickness difference discrimination by an echolocating Atlantic bottlenose dolphin
,”
J. Comp. Physiol. A
170
,
41
47
.
8.
Aubauer
,
R.
, and
Au
,
W. W. L.
(
1998
). “
Phantom echo generation: A new technique for investigating dolphin echolocation
,”
J. Acoust. Soc. Am.
104
,
1165
1170
.
9.
Aubauer
,
R.
,
Au
,
W. W. L.
,
Nachtigall
,
P. E.
,
Pawloski
,
D. A.
, and
DeLong
,
C. M.
(
2000
). “
Classification of electronically generated phantom targets by an Atlantic bottlenose dolphin (Tursiops truncatus)
,”
J. Acoust. Soc. Am.
107
,
2750
2754
.
10.
Bilsen
,
F. A.
, and
Ritsma
,
R. J.
(
1969
). “
Repetition pitch and its implication for hearing theory
,”
Acta Acust. Acust.
22
(
2
),
63
73
.
11.
Branstetter
,
B. K.
,
Van Alstyne
,
K. R.
,
Strahan
,
M. G.
,
Tormey
,
M. N.
,
Wu
,
T.
,
Breitenstein
,
R. A.
,
Houser
,
D. S.
,
Finneran
,
J. J.
, and
Xitco
,
M. J.
, Jr.
(
2020
). “
Spectral cues and temporal integration during cylinder echo discrimination by bottlenose dolphins (Tursiops truncatus)
,”
J. Acoust. Soc. Am.
148
(
2
),
614
626
.
12.
Dubrovsky
,
N. A.
(
1990
). “
On the two auditory subsystems in dolphins
,” in
Sensory Abilities of Cetaceans: Laboratory and Field Evidence
, edited by
J. A.
Thomas
and
R. A.
Kastelein
(
Plenum
,
New York
), pp.
233
254
.
13.
Dubrovsky
,
N. A.
,
Zorikov
,
T. V.
,
Kvighinadze
,
O. S.
, and
Kuratishvili
,
M. M.
(
1992
). “
Mechanisms of signal discrimination and identification in the auditory system of Tursiops truncatus
,” in
Marine Mammal Sensory Systems
, edited by
J. A.
Thomas
,
R. A.
Kastelein
, and
A. Y.
Supin
(
Plenum
,
New York
), pp.
235
240
.
14.
Finneran
,
A. E.
,
Mulsow
,
J.
, and
Finneran
,
J. J.
(
2020a
). “
Detection of simulated patterned echo packets by bottlenose dolphins (Tursiops truncatus)
,”
J. Acoust. Soc. Am.
148
,
1007
1013
.
15.
Finneran
,
J. J.
,
Jones
,
R.
,
Guazzo
,
R. A.
,
Strahan
,
M. G.
,
Mulsow
,
J.
,
Houser
,
D. S.
,
Branstetter
,
B. K.
, and
Moore
,
P. W.
(
2020b
). “
Dolphin echo-delay resolution measured with a jittered-echo paradigm
,”
J. Acoust. Soc. Am.
148
,
374
388
.
16.
Hammer
,
C. E.
, Jr.
, and
Au
,
W. W. L.
(
1980
). “
Porpoise echo-recognition: An analysis of controlling target characteristics
,”
J. Acoust. Soc. Am
68
,
1285
1293
.
17.
Helweg
,
D. A.
,
Moore
,
P. W.
,
Dankiewicz
,
L. A.
,
Zafran
,
J. M.
, and
Brill
,
R. L.
(
2003
). “
Discrimination of complex synthetic echoes by an echolocating bottlenose dolphin
,”
J. Acoust. Soc. Am.
113
,
1138
1144
.
18.
Ming
,
C.
,
Haro
,
S.
,
Simmons
,
A. M.
, and
Simmons
,
J. A.
(
2021
). “
A comprehensive computational model of animal biosonar signal processing
,”
PLoS Comput. Biol.
17
,
e1008677
.
19.
Mogdans
,
J.
,
Schnitzler
,
H. U.
, and
Ostwald
,
J.
(
1993
). “
Discrimination of two-wavefront echoes by the big brown bat, Eptesicus fuscus: Behavioral experiments and receiver simulations
,”
J. Comp. Physiol. A
172
,
309
323
.
20.
Moore
,
P. W. B.
,
Hall
,
R. W.
,
Friedl
,
W. A.
, and
Nachtigall
,
P. E.
(
1984
). “
The critical interval in dolphin echolocation: What is it?
,”
J. Acoust. Soc. Am.
76
,
314
317
.
21.
Mulsow
,
J.
(
2022
). (private communication).
22.
Pailhas
,
Y.
,
Capus
,
C.
, and
Brown
,
K.
(
2012
). “
Dolphin-inspired sonar system and its performance
,” in
IET Radar, Sonar, Navig.
6
(
8
),
753
763
.
23.
Popov
,
V. V.
, and
Supin
,
A.
(
1997
). “
Detection of temporal gaps in noise in dolphins: Evoked-potential study
,”
J. Acoust. Soc. Am.
102
,
1169
1176
.
24.
Popov
,
V. V.
, and
Supin
,
A. Y.
(
1998
). “
Auditory evoked responses to rhythmic sound pulses in dolphins
,”
J. Comp. Physiol. A
183
,
519
524
.
25.
Popov
,
V. V.
,
Supin
,
A. Y.
, and
Klishin
,
V. O.
(
2001
). “
Auditory brainstem response recovery in the dolphin as revealed by double sound pulses of different frequencies
,”
J. Acoust. Soc. Am.
110
,
2227
2233
.
26.
Saillant
,
P. A.
,
Simmons
,
J. A.
,
Dear
,
S. P.
, and
McMullen
,
T. A.
(
1993
). “
A computational model of echo processing and acoustic imaging in frequency-modulated echolocating bats: The spectrogram correlation and transformation receiver
,”
J. Acoust. Soc. Am.
94
,
2691
2712
.
27.
Sanderson
,
M. I.
, and
Simmons
,
J. A.
(
2000
). “
Neural responses to overlapping FM sounds in the inferior colliculus of echolocating bats
,”
J. Neurophysiol.
83
,
1840
1855
.
28.
Shriram
,
U.
, and
Simmons
,
J. A.
(
2019
). “
Echolocating bats perceive natural-size targets as a unitary class using micro-spectral ripples in echoes
,”
Behav. Neurosci.
133
,
297
304
.
29.
Simmons
,
J. A.
,
Freedman
,
E. G.
,
Stevenson
,
S. B.
,
Chen
,
L.
, and
Wohlgemant
,
T. J.
(
1988
).
Clutter Interference and the Integration Time of Echoes in the Echolocating Bat, Eptesicus fuscus
(
Brown University Press
,
Providence, RI
).
30.
Simmons
,
J. A.
,
Houser
,
D.
, and
Kloepper
,
L.
(
2014
). “
Localization and classification of targets by echolocating bats and dolphins
,” in
Biosonar
, edited by
A.
Surlykke
,
P. E.
Nachtigall
,
R. R.
Fay
, and
A. N.
Popper
(
Springer
,
New York
), pp.
169
193
.
31.
Simmons
,
J. A.
,
Moss
,
C. F.
, and
Ferragamo
,
M.
(
1990
). “
Convergence of temporal and spectral information into acoustic images of complex sonar targets perceived by the echolocating bat, Eptesicus fuscus
,”
J. Comp. Physiol. A
166
,
449
470
.
32.
Strahan
,
M. G.
,
Finneran
,
J. J.
,
Mulsow
,
J.
, and
Houser
,
D. S.
(
2020
). “
Effects of dolphin hearing bandwidth on biosonar click emissions
,”
J. Acoust. Soc. Am.
148
,
243
252
.
33.
Supin
,
A. Y.
, and
Popov
,
V. V.
(
1995a
). “
Envelope-following response and modulation transfer function in the dolphin's auditory system
,”
Hear. Res.
92
,
38
46
.
34.
Supin
,
A. Y.
, and
Popov
,
V. V.
(
1995b
). “
Temporal resolution in the dolphin's auditory system revealed by double-click evoked potential study
,”
J. Acoust. Soc. Am.
97
,
2586
2593
.
35.
Thurlow
,
W. R.
, and
Small
,
A. M.
(
1955
). “
Pitch perception for certain periodic auditory stimuli
,”
J. Acoust. Soc. Am.
27
,
132
137
.
36.
Vel'min
,
V. A.
, and
Dubrovsky
,
N. A.
(
1976a
). “
Auditory analysis of sound pulses in dolphins
,”
Dokl. Akad. Nauk SSSR
225
,
470
473
.
37.
Vel'min
,
V. A.
, and
Dubrovsky
,
N. A.
(
1976b
). “
The critical interval of active hearing in dolphins
,”
Sov. Phys. Acoust.
2
,
351
352
.
38.
Yost
,
W. A.
, and
Hill
,
R.
(
1978
). “
Strength of the pitches associated with ripple noise
,”
J. Acoust. Soc. Am.
64
,
485
492
.
39.
Zorikov
,
T. V.
, and
Dubrovsky
,
N. A.
(
2003
). “
Echo-processing in bottlenose dolphins
,” in
Oceans 2003 Conference
,
San Diego, CA
, pp.
320
326
.