Short-term enhancement and suppression of dolphin auditory evoked responses following echolocation click emission

A key feature of sonar systems is the ability to compensate for changes in echo strength that naturally occur with changes in target range, so that targets with the same size and reflective characteristics produce echoes with similar strength, regardless of range. For hardware sonars, this is commonly accomplished using time-varying gain, where the receiver sensitivity increases with time after pulse emission to compensate for the range-dependent transmission loss. For echolocating animals, the term automatic gain control (AGC) has been used to describe processes by which animals compensate for changes in echo strength with target range (Kick and Simmons, 1984; Au and Benoit-Bird, 2003). For odontocetes (toothed whales), three AGC mechanisms have been identified: (1) changes to the transmitted sonar pulse (“click”) amplitude with target range (e.g., Rasmussen 2002; Au and Benoit-Bird, 2003; Au, 2004; Jensen 2009), (2) changes in the subject's hearing sensitivity during echolocation (e.g., Supin 2006, 2008a), and (3) the progressive release of the neuronal response to the echo from forward masking by the outgoing pulse as target range increases (e.g., Supin 2008b, 2009).

Changes in hearing sensitivity during echolocation and the release of echoes from forward masking have been studied by measuring auditory evoked potentials (AEPs) while animals echolocate (Supin 2003). Most studies have examined the relationships between click level, echo level, and target range (echo delay) and the transient AEPs evoked by the animal's emitted click (the click-evoked potential, or clickEP) and by the returning echo (the echo-evoked potential, or echoEP) (e.g., Supin 2003; Supin 2009; Li 2011; Linnenschmidt 2012). Studies have shown that, as target range increases up to 10 to 20 m, the progressive release of the echo from forward masking by the click counteracts the decrease in acoustic echo strength that occurs with range, keeping the amplitudes of neuronal responses to echoes approximately constant despite large changes in acoustic echo strength (Supin 2005; Finneran 2013b).

In a slightly different approach, Supin (2008a) examined the evoked response to an external sound stimulus while a false killer whale (Pseudorca crassidens) performed an echolocation task. External stimuli consisted of sinusoidal amplitude modulated (SAM) tone bursts, which elicit the auditory steady-state response (ASSR). The ASSR is formed from a succession of overlapping evoked potentials and possesses a fundamental frequency at the SAM tone modulation rate (Picton 2003). Supin (2008a) reported changes in the ASSR amplitude as a function of target strength and target present/absent condition; however, since the SAM tone bursts were presented independently of the animal's echolocation pulse, changes in ASSR amplitude could only be examined at the level of a trial (i.e., across multiple clicks) and not within the temporal scale of an individual click-echo pair.

Finneran (2013a) combined the external sound stimulation approach of Supin (2008a) with a moving average analysis similar to that employed in swept-parameter ASSR studies (e.g., Picton 2007; Finneran 2011) to allow the ASSR amplitude induced by a 113-kHz SAM tone to be observed on time scales relevant to a single click-echo pair. The results showed a rapid decrease in ASSR amplitude at the time of click emission, followed by a 25 to 75 ms recovery (Finneran 2013a); however, the results were limited to only a single frequency and thus provided no indication as to the frequency extent of the observed suppression in the ASSR. Further, the specific analysis performed resulted in a temporal ambiguity of ±2 ms, obscuring the temporal relationship between click emission and ASSR suppression and making it difficult to unequivocally rule out a mechanism such as the middle ear reflex that would begin to suppress the ASSR before the click was emitted.

The present study expands on the work of Finneran (2013a) by utilizing SAM tone frequencies from 25 to 125 kHz. The goal was to examine the effects of the emitted click on the dolphin's hearing ability as a function of frequency and time relative to click emission. The basic methodology followed that of the previous study: ASSR amplitudes were measured in response to continuous SAM tones while a bottlenose dolphin (Tursiops truncatus) performed an echolocation task. Rather than physical targets, the echolocation task utilized a phantom echo generator (PEG) to produce echoes by convolving the dolphin's emitted click with the impulse response of a simulated target, then broadcasting the delayed echo waveforms to the dolphin. The results showed not only suppression and recovery of the ASSR following click emission, but also, at the lower frequencies, a brief enhancement of the ASSR after click emission.

The subject was a male Atlantic bottlenose dolphin (TRO), 21-yr old, with a high-frequency hearing limit between 120 and 140 kHz. The test site was identical to that used with TRO during previous phantom echo studies (e.g., Finneran 2013b). Ambient noise at the site was dominated by snapping shrimp, other dolphins, and vessel traffic. The average noise pressure spectral density from all trials, over a frequency range corresponding to the mean center frequency ± one-half the rms bandwidth (Au, 1993) of TRO's echolocation clicks, was 54 dB re 1 μPa²/Hz.

Test sessions were conducted in a 9 m × 9 m floating netted enclosure containing an observation aperture facing San Diego Bay [Fig. 1(a)]. The aperture was a 1.8-m square frame constructed of polyvinylchloride (pvc) pipe and covered with netting except for a 35-cm diameter hoop opening in the center, at a depth of 1 m. The dolphin was trained to position his head in the hoop facing outward towards San Diego Bay. Several 50-cm concrete piles and a single 46-cm diameter hollow steel pile were located in front of the hoop at various distances and azimuths [see Fig. 1(b)]; otherwise, the water in front of the hoop was free from man-made structures to a distance of 1 km. The presence of the piles did not appear to have a significant effect on TRO's performance or click emissions compared to a previous task in an uncluttered environment (Finneran, 2013).

Trials were differentiated by whether the dolphin was performing an echolocation task (echolocation trials) and whether external SAM tones were presented to the dolphin (SAM tone trials). The condition with no echolocation and no SAM tones was of no interest, therefore there were three trial types over which data were collected: (1) echolocation trials with SAM tones (experimental trials), (2) echolocation trials without SAM tones (“no SAM” control trials), and (3) SAM tone trials with no echolocation (“no PEG” control trials).

During echolocation trials, TRO positioned himself in the hoop, produced echolocation clicks and listened to the returning phantom echoes. Two biosonar target impulse responses, denoted as A and B (Finneran 2010; Finneran 2013b), were used. TRO was trained to produce an echolocation burst pulse when he detected a change from echo A to echo B, and to withhold the burst pulse otherwise. Eighty-percent of the trials were echo change (EC) trials, where the target impulse response changed from target A to target B after a random interval of 5 to 10 s followed by a 2-s response interval. The remainder of the trials were no change (NC) trials, where the impulse response remained at target A for the 7- to 12-s trial duration. If the dolphin responded during the 2-s response interval after a target change (a hit), or withheld the response for an entire NC trial (a correct rejection), it was rewarded with one fish. A response outside of a response interval (a false alarm) or a failure to respond during a response interval (a miss) resulted in the dolphin being recalled to the surface with no fish reward. A response before the echo change during an EC trial was classified as a false alarm and the trial was therefore considered to be an NC trial rather than an EC trial. If the dolphin did not echolocate during a trial, stopped echolocating before the echo change, left the hoop, or was visually observed to be echolocating on another object, he was recalled and the trial data were discarded. Echolocation trials were conducted with simulated target ranges of 25 and 50 m. During “no PEG” trials, TRO positioned himself next to the hoop, on a “biteplate,” while remaining quiet (i.e., not producing echolocation clicks).

During SAM tone trials, a continuous SAM tone was presented to the dolphin via “jawphone” contact transducers. A number of different SAM tone frequencies and levels were utilized (see below and Table I). During “no SAM” controls, the SAM tone amplitude was set to zero volts, but the dolphin still wore the sound projectors (see below).

Each experimental session lasted approximately 60 to 90 min and consisted of 120 discrete trials. Sessions were divided into blocks of 5 trials within which the trial type and conditions (i.e., target range and/or SAM tone level and frequency) were fixed. Sessions typically began and ended with a block of no SAM controls. The remainder of the trials were divided into groups of 25 to 30 trials within which the SAM tone frequency was fixed. Within these groups, blocks alternated between no PEG controls, echolocation at 25-m range, and echolocation at 50-m range. SAM tone frequency and sound pressure level (SPL) were randomized across test days.

The PEG used in this experiment has been previously described in detail (Finneran 2010; Finneran, 2013). Two piezoelectric transducers (1089D, International Transducer Corp, Santa Barbara, CA and TC4013, Reson Inc., Slangerup, Denmark) were located 1 m from the subject's blowhole when positioned in the hoop. Phantom echoes were produced by capturing the dolphin's outgoing echolocation clicks with the 1089D and convolving the clicks with the impulse response of a target to create echo waveforms. Analog echo waveforms were filtered (5–200 kHz, 3 C module, Krohn-Hite Corporation, Brockton, MA), attenuated (PA5, Tucker-Davis Technologies, Alachua, FL), amplified (CC4000, Crest Audio, Meridian, MS), and broadcast to the dolphin using the TC4013. Simulated target ranges were either 25 or 50 m. The relative echo levels (RELs)—the amplitudes of echoes relative to the click source level—were independent of simulated range but varied from trial to trial (see below), with the REL of echo A between −65 and −85 dB, depending on simulated range, SAM tone frequency, and SAM tone SPL. The REL of echo B was always 5 dB lower than that of echo A.

Previous studies have shown that echoEP and ASSR amplitudes during echolocation are heavily influenced by click peak-peak (p-p) SPL and echo delay, and that click SPLs will vary from trial-to-trial, even if echo delay and REL are constant (e.g., Supin 2009; Finneran 2013b). Pilot data for the present study showed that click p-p SPLs were generally between 200 and 210 dB re 1 μPa, but varied substantially with simulated range, SAM tone SPL, and SAM tone frequency. Therefore, to reduce confounds associated with changing click SPLs while maximizing the amount of data that could be analyzed, the experimenter attempted to keep click p-p SPLs between 200 and 210 dB re 1 μPa by simultaneously adjusting the RELs of echo A and B (using the PA5 attenuator) during each trial while monitoring the emitted click SPL (as echo attenuation increased, TRO tended to produce higher-SPL clicks).

A third piezoelectric transducer (TC4013, Reson Inc., Slangerup, Denmark) was embedded in a silicon suction cup and placed on the dolphin's melon. This was done to obtain a measure of click amplitude independent of head movement. During data analysis, the signal recorded from this “melon hydrophone” was compared to the farfield signal recorded from the ITC1089D to ensure that data associated with echolocation clicks directed away from the farfield hydrophone were not included in the analysis. The presence of the melon hydrophone did not appear to have a significant effect on the subject's echolocation emissions, performance, or the farfield click recordings.

Auditory evoked potentials were measured using surface electrodes embedded in suction cups and attached to the dolphin's head and back. A small amount of conductive paste was applied to the electrodes just prior to attachment. A biopotential amplifier (ICP511; Grass Technologies, West Warwick, RI) filtered (300–3000 Hz) and amplified (100 dB) the potential difference between an electrode located on the midline, approximately 5 cm from the posterior edge of the blowhole, and one placed near the right external auditory meatus (a third, common electrode was placed near the dorsal fin). The differential electrode voltage, representing the instantaneous electroencephalogram (EEG), was digitized using a PXIe 6368 data acquisition device (National Instruments, Austin, TX). Two synchronization signals were also generated and recorded along with the instantaneous EEG: the “echo sync pulse” was a rectangular pulse time-locked with the echo waveform and the “ASSR sync pulse” was a 1-kHz square wave, phase-locked to the SAM tone modulation. Finally, signals from the PEG farfield hydrophone and melon hydrophone were also digitized using the PXIe 6368. All analog-digital conversions in the PXIe 6368 were performed with a 2-MHz sampling rate and 16-bit resolution. Each trial therefore resulted in a single WAV file containing the (1) farfield hydrophone signal, (2) echo sync pulse, (3) melon hydrophone signal, (4) instantaneous EEG, and (5) ASSR sync pulse.

External sound stimuli for the ASSR measurements consisted of continuous SAM tones with a modulation frequency of 1 kHz and depth of 100%. Stimuli were digitally synthesized, then converted to analog with 16-bit resolution and 2-MHz sample rate using a second PXIe 6368 data acquisition device. Analog signals were filtered (5–200 kHz, 3 C module, Krohn-Hite Corporation, Brockton, MA), amplified (7600M, Krohn-Hite Corporation, Brockton, MA), then projected via “jawphone” transducers consisting of TC4013s embedded in silicon suction cups and attached to the dolphin's left and right lower jaws [Fig. 1(a)]. Data were collected at SAM tone frequencies of 25, 32, 40, 50, 63, 80, 100, and 125 kHz. During preliminary data collection, ASSR amplitudes were measured at these frequencies when TRO was not echolocating. The resulting ASSR amplitude curves (Fig. 2) were used to identify the appropriate transducer excitation voltages for use in the main experiments. The absence of electrical artifacts in the instantaneous EEG at each SAM tone frequency and SPL was verified from periodic measurements utilizing an intermittent stimulus, so that the latency of the auditory evoked response could be observed.

ASSR data collection during the echolocation task occurred in two phases. During initial testing, transducer electrical driving voltages of 5 dBV (0-peak) were used for frequencies above 32 kHz, while 10 dBV was used at 25 and 32 kHz (where ASSR amplitudes were smaller). Data collection was then repeated at frequencies from 32 to 100 kHz, with the driving voltage reduced as much as possible while still obtaining measurable ASSRs during echolocation. (Reducing driving voltages at 25 and 125 kHz resulted in ASSR amplitudes too small for analysis.) In this paper, data are presented in two groups: Group 1 consists of data obtained with SAM tone frequencies of 25, 125 kHz, and the lower amplitude 32, 40, 50, 63, 80, and 100 kHz tones. Group 2 consists of data collected with the higher amplitude SAM tones from 32 to 100 kHz.

The dolphin's performance in the echolocation task was quantified using the hit rate (the number of hits divided by the number of EC trials) and the false alarm rate (the number of false alarms divided by the number of NC trials).

Initial EEG data analysis consisted of examining the echo sync pulse recording from each trial to identify individual, 250-ms time “epochs” in the instantaneous EEG signal beginning 50 ms before each click was produced by the dolphin. For each time epoch, the click p-p SPL, center frequency, rms bandwidth, and inter-click interval (ICI) were computed using custom software. Next, the instantaneous EEG and ASSR sync pulse tracks for each time epoch were extracted, down-sampled to 40-kHz, and saved as a separate file (an EEG waveform “clip”) that was coded with the corresponding echo (A or B), range, and click p-p SPL. During clip extraction, the EEG clips were high-pass filtered using a digital implementation of a zero-phase shift, 8th-order Butterworth filter with a 200-Hz cutoff frequency. Time intervals in the original trials containing a burst pulse response, having an inter-click interval less than the time between emission of a click and reception of the echo (the two-way travel time), or containing echoes corresponding to target B were excluded from EEG clip extraction. Changes in click p-p SPL observed over the course of each trial were also compared to click amplitudes obtained from the melon hydrophone to determine if changes in click SPL were a result of head movement; if so, the affected time segments were also excluded from EEG clip extraction.

For each experimental condition (range, SAM tone frequency/level, or control), the number of available EEG clips was computed for various click p-p SPL bins to find the smallest range of click SPLs for which at least 2000 EEG clips were available for all conditions. This resulted in an SPL range of 206.5 ± 3 dB re 1 μPa (204 to 209 dB re 1 μPa). Table I shows the number of EEG clips available for each experimental condition.

Previous measurements of ASSR amplitudes during echolocation showed that the presence of the click-EP and echo-EP could interfere with interpretation of the ASSR (Finneran 2013a). Therefore, within each experimental condition, the EEG clips were first synchronously averaged with the clips aligned with the emitted click [Fig. 3(a)]; this improved the signal-to-noise ratio of the clickEP and echoEP by reducing residual background noise that was not synchronized with the dolphin's clicks, which included the ASSR. The resulting averaged EEG was then used as a “template” and subtracted from each individual EEG clip from the same experimental condition to eliminate the clickEP and echoEP waveforms from each epoch. After this subtraction, the EEG clips were temporally aligned by “shifting” the start of each clip to match the onset of the nearest cycle of the ASSR sync pulse. This process resulted in some ambiguity in the temporal alignment between the ASSR and the click, up to ±0.5 ms (i.e., one-half the period of the 1-kHz modulation signal). After clip alignment with the ASSR sync pulse, synchronous averaging was performed in the time domain to obtain the ASSR time waveform [Fig. 3(b)]. A 5-ms, rectangular window was also moved across each 250-ms clip, with 50% overlap between successive window temporal positions (i.e., the window was moved in 2.5-ms steps). At each window position, a Fourier transform was performed and the complex spectral amplitude at the SAM tone modulation rate (1 kHz) was used to calculate the ASSR amplitude [Fig. 3(c)] and the magnitude squared coherence (MSC) based on 16 “subaverages” (Dobie and Wilson, 1996). The MSC statistic was compared to a critical value at the level of α = 0.01 to test if the ASSR was significantly different from the residual background noise in the EEG. Finally, the ASSR data for each condition [e.g., Fig. 3(c)] were divided by the mean of the ASSR amplitudes measured in the absence of echolocation [the “control” condition, e.g., Fig. 3(d)] to obtain normalized ASSR amplitudes as a function of time relative to click emission [Fig. 3(e)], for each combination of range, SAM tone frequency, and SAM tone SPL.

A total of 270 control and 1038 echolocation trials were conducted, with 34 to 87 trials at each target range condition, which resulted in 2060 to 5635 individual EEG clips, depending on the range and SAM tone characteristics (Table I). Hit rates varied from 0.96 to 1.0, while false alarm rates varied from 0.00 to 0.25, depending on the experimental conditions. Increases in false alarm rates while maintaining high detection rates were previously reported for dolphins performing similar echo change-detection tasks (Finneran, 2013). Higher false alarm rates tended to occur with SAM tone frequencies from 50 to 100 kHz, suggesting that the SAM tones were partially masking the echoes.

Click properties for the various experimental conditions are summarized in Fig. 4. Within the 204 to 209 dB re 1 μPa range of clicks included in the study, center frequencies and rms bandwidths varied between ∼70 to 95 kHz and 30 to 35 kHz, respectively, for all conditions. The only systematic trends were seen in the Group 2 (higher SAM tone SPL) data, where click p-p SPLs increased in the presence of the SAM tones and the click center frequencies tended to be lower (mean of 78–80 kHz compared to 83–85 kHz) in the presence of the 80 and 100 kHz tones compared to the 32 to 63 kHz tones. This suggests that the dolphin may have manipulated click emissions to reduce masking effects of the SAM tone on echo reception. Inter-click intervals were broadly distributed for 25 -m range but more narrowly distributed at 50 m; the tight distributions at 50 m were likely a result of excluding data with ICIs less than the two-way travel time (67 ms for 50 -m range) coupled with the natural tendency for bottlenose dolphins to limit ICIs to less than ∼100 ms (Finneran, 2013). Mean click spectra (Fig. 5) were essentially flat from ∼40 to 115 kHz, regardless of experimental condition.

When the normalized ASSR amplitudes are viewed as functions of time (Figs. 6 and 7), two main patterns emerge across the experimental conditions: When SAM tone frequencies were above 40 to 50 kHz, ASSR amplitudes were suppressed just after click emission, then recovered over a period of tens of milliseconds, typically returning to the pre-click baseline value sometime before the echo. However, for SAM tone frequencies at and below 40 to 50 kHz, an enhancement of the ASSR amplitude occurred 5 to 10 ms after click emission. In most cases, the enhancement was preceded by a brief (∼5 ms) suppression. The time course for enhancement was similar to that of suppression, and was normally less than the two-way travel time. For Group 1 data, enhancement occurred at 50 kHz and below. For Group 2 data (larger SAM tone amplitudes), enhancement occurred at 40 kHz and below, and ASSR amplitudes at 50 kHz exhibited suppression rather than enhancement. The enhancement phase lasted up to 50 to 60 ms, during which ASSR amplitudes increased by as much as 70%. During ASSR suppression, amplitudes dropped as low as 14% of the control amplitude. Finally, for some experimental conditions, baseline ASSR amplitudes during echolocation were smaller than those observed during control sessions with no echolocation, indicating incomplete recovery of the ASSR before the emission of the next click and a residual suppressive effect from one click to the next. This effect was most pronounced at 80 kHz, near the center frequency of the emitted click, and with the lower amplitude SAM tones (Group 1).

The present study expands upon previous work examining the feasibility of measuring the effects of biosonar click emission on ASSR amplitudes at 113 kHz (Finneran 2013a). Limitations of the previous study included the ±2 ms time ambiguity introduced by “jittering” EEG clip start times before averaging and the 10-ms time window used for spectral analysis. The jittering was introduced to reduce the clickEP amplitude within the grand average EEG and allow the ASSR to be obtained, while the 10-ms spectral window was necessary to lower residual background noise within spectral bins to achieve sufficient ASSR signal-to-noise ratio. In the present study, the potential for the clickEP to obscure the ASSR was minimized by first subtracting a template AEP, obtained by synchronously averaging EEG clips aligned with the biosonar click from each EEG clip. The EEG clips were then aligned with the ASSR sync pulse and synchronously averaged. This process was successful in eliminating the clickEP and echoEP from the ASSR grand average, while introducing only a ±0.5 ms time ambiguity (one-half the SAM tone modulation period). The present study also featured a 5-ms spectral analysis window, further reducing the temporal ambiguity associated with the ASSR amplitude-versus-time data (Figs. 6 and 7). While further reduction of the spectral window size is desirable from the standpoint of improving temporal resolution in the ASSR amplitude data, the concomitant drop in ASSR signal-to-noise limits the minimum practical spectral window size for small ASSR amplitudes. Finally, data were analyzed for a relatively small (±3 dB) range of click p-p SPL, allowing the effects of click SPL to be largely controlled.

The main findings of the present study relate to the changes in ASSR amplitude observed over the time course of a click-echo pair. At frequencies above 40 to 50 kHz, there was an overall, small suppression in ASSR amplitudes compared to the control condition, plus a larger, temporary suppression after click emission. At lower frequencies, ASSR amplitudes were temporarily enhanced after click emission. Based on previous work with bats and odontocetes, possible mechanisms for the temporary reduction and recovery of response amplitude, which presumably indicates a receiver-based form of gain control, could include a middle ear reflex, release from forward masking, or a change in hearing sensitivity related to echolocation activity (Henson, 1965; Supin and Nachtigall, 2013).

Suppression of hearing following click emission has been previously reported for both echolocating bats and odontocetes. Bats contract the stapedius muscle 4 to 10 ms before the emission of a sonar pulse and relax the muscle afterwards (Henson, 1965), attenuating the sound transmitted through the middle ear complex and lowering hearing sensitivity (Kick and Simmons, 1984). In contrast, the primary mechanism identified for receiver-based gain control in odontocetes is the release from forward masking by the emitted click: studies have shown that the emitted click can mask returning echoes, with the extent and duration of the forward masking effect depending on the relative amplitudes and temporal spacing between the click and echo (Supin 2009, 2011).

For the present study, interpretation of the temporal patterns of ASSR suppression (Fig. 6) are complicated by the spectral window size (5 ms) and the latency of the ASSR in dolphins measured using a 1 kHz modulation rate (∼4 ms). Examination of Fig. 6, taking into account the spectral window size, indicates no appreciable change in ASSR amplitude up to 3 ms after click emission, therefore the events responsible for suppression of the ASSR occur close in time (i.e., within roughly a millisecond) to click emission. This timing relationship and the relatively slow time course of the recovery from suppression, requiring up to 30 to 60 ms, suggest a forward masking-based mechanism rather than a middle-ear reflex (Henson, 1965). The frequency pattern of suppression, where frequencies closer to the click central frequency bandwidth are suppressed more strongly, also fits the behavior expected from forward masking better than that expected from stapedial contraction (Suga and Jen, 1975; Geisler, 1998; Pickles, 2008).

The overall reductions in ASSR amplitudes (relative to control levels) near click center frequencies are likely related to a residual masking effect from the continuous emission of clicks during echolocation. This demonstrates that, under certain conditions, it is possible for a dolphin to experience an overall suppression of auditory system responsiveness (in addition to temporary suppression/recovery) while emitting echolocation clicks. This reduction likely depends on both the SPL of the outgoing clicks and the ICI. The limited range of SPLs and ICIs (i.e., ranges) in the current study cannot adequately describe this relationship, however, and it is not clear if this phenomenon imposes any functional limitations on the echolocation system under natural conditions.

Enhancement of brainstem responses following biosonar click emissions (Figs. 6 and 7) has not been previously reported, and the underlying mechanisms are unclear. Although forward masking is typically associated with a reduction in response amplitude, a number of psychophysical and electrophysiological studies have also reported enhancement effects (reviewed by Henry, 1991a). Henry (1991a,b) found that forward masking could increase the amplitude of the compound action potential (CAP) in gerbils under certain tone-on-tone forward masking conditions. Enhancement effects, defined as a CAP amplitude at least 20% larger than that in the unmasked condition, were seen in roughly 2/3 of the animals tested. The most significant enhancement occurred with low-level maskers near the probe tone frequency and, less commonly, higher-level maskers below the probe frequency. In contrast, enhancement resulting from a masker with a frequency above that of the probe was only observed in 1 out of 32 subjects (Henry, 1991a). The frequency patterns of the enhancement observed by Henry (1991a,b) therefore do not match those of the present study, where enhancement was seen only within the lower frequency band of the masker and most of the masker energy was above the probe frequency. Enhancement of human auditory brainstem responses has also been found during tone-on-tone forward masking experiments (Ananthanarayan and Gerken, 1983, 1987). Ananthanarayan and Gerken (1983, 1987) found that a forward masker with frequency near the probe frequency suppressed wave III amplitude but enhanced wave V amplitude. The ASSR in dolphins primarily reflects the dolphin P4-N5 complex (Popov and Supin, 1990) and may be homologous to the human wave V (Bullock 1968; Ridgway 1981); however, once again the frequency pattern of enhancement did not follow the pattern observed in the present study. Therefore, although terrestrial mammal electrophysiological studies have demonstrated response enhancement during forward masking conditions, masker and probe conditions have differed and the observed frequency patterns did not match the results of the present experiment. Furthermore, terrestrial mammal studies have not clearly identified specific mechanisms responsible for response enhancement, though Ananthanarayan and Gerken (1987) suggested a central, rather than peripheral, mechanism for wave V enhancement.

It is possible that the brainstem response enhancement observed in the present study is the result of an active enhancement of sensitivity during echolocation. Supin (2008a) reported changes in the ASSR to a 22-kHz SAM tone in a false killer whale performing a physical target detection task, with ASSR amplitudes higher during target absent trials compared to target present trials. Nachtigall and Supin (2015) also reported a conditioned reduction in ASSR amplitude in a dolphin when an intense sound was preceded by warning sound. The ASSR amplitude reduction occurred at frequencies equal to or higher than the intense tone frequency. These studies demonstrate that manipulation of hearing sensitivity is possible in odontocetes under certain conditions. This raises the possibility that the ASSR patterns observed in the present study reflect an overall boost in auditory system responsiveness—across a broad frequency range—coupled with a suppression of the ASSR at the higher frequencies due to forward masking. However, it is not known if manipulation of hearing sensitivity in this fashion could occur within the timeframe of a single click-echo pair. The underlying mechanisms through which this could occur are also unknown. Nachtigall and Supin (2015) speculated that the frequency-dependent conditioned suppression may utilize feedback circuits involving outer hair cells. One of the interesting patterns in the current data that may aid in determining the nature of this mechanism (or mechanisms) is the fact that, although similar levels of ASSR suppression and enhancement relative to controls were observed, there were no residual overall enhancement effects comparable to those observed with suppression.

It is also possible that the brainstem response enhancement is simply a result of conventional masking (i.e., not related to active echolocation). Off-frequency masking related to lateral suppression is a well-known feature of the mammalian auditory system, and complex masking patterns related to lateral suppression have been documented in dolphins (Popov 1997, 1998). Suppressing tone bursts result in a “rebound” of activity in auditory nerve fibers when the suppressor is turned off (Pickles, 2008); however, it is not known if such a “release” from lateral suppression could result in the response enhancement patterns observed in the present study.

Additional data are required to understand the origins of and relationships between the suppression and enhancement effects seen in the present study. Experiments using simulated clicks rather than the dolphin's own biosonar emissions would shed light on whether the observed relationships are unique to active echolocation processes. The use of tone burst probe stimuli, rather than a continuous SAM tone, would enable hearing evaluation at specific times and frequencies and allow the effects on individual brainstem response components to be assessed (i.e., early, peripheral vs later, more central processing).

Finally, the early deflections observed in the AEP templates obtained after averaging EEG clips aligned with the biosonar click [Fig. 3(a), Fig. 8] are worth noting. These waves begin near the time of click emission and overlap the early waves in the clickEP. The origin of these signals is unknown—not only are the latencies too short to indicate auditory nerve or brainstem activity, but the deflections actually begin several hundred microseconds before click emission. Measurements made while manipulating the distance between the blowhole and the nearest electrode showed that increasing the caudal distance between the blowhole and the electrode reduced the amplitudes of these waves. This suggests that the early deflections in the averaged EEG were artifacts caused by proximity to the click generator and not electrical artifacts in the recording system itself. These waves may be myogenic potentials related to the click generation apparatus. Ridgway (1980) measured electromyograms (EMGs) from several muscles in the nasal system and reported bursts of activity associated with click production. Henson (1965) measured EMGs associated with stapedius muscle contraction in bats preceding click emissions at click rates up to 100 clicks/s. Together, these data suggest that it may be possible for muscle contraction associated with individual click generation to occur in the nasal system in dolphins, within the time-frame relevant to the inter-click intervals observed in the present study (∼40 to 100 ms). However, the time-course of the early waves measured in the present study (∼1 ms) is shorter than those typically seen in farfield EMG recording; therefore, if these waves truly represent an EMG related to click production, the underlying muscle activity must possess a very high-degree of synchrony.

The authors thank Megan Tormey and Arial Brewer for dolphin training and handling during the experimental sessions and Randall Dear and Jim Powell for logistic support. The study followed a protocol approved by the Institutional Animal Care and Use Committee at the Biosciences Division, Space and Naval Warfare Systems Center (SSC), Pacific and the Navy Bureau of Medicine and Surgery, and followed all applicable U.S. Department of Defense guidelines. Financial support was provided by the SSC Pacific Naval Innovative Science and Engineering (NISE) program and the Office of Naval Research Code 32 (Mine Countermeasures, Acoustics Phenomenology & Modeling Group).