Development of behavioral observation audiometry tests for companion dogs

Abstract

Behavioral observational audiometry is applied to study the hearing function of preverbal children aged less than 6 months. We hypothesized that this method could be used to assess hearing threshold in untrained companion dogs. The small number of studies that examine untrained behavioral responses to sounds have not described determination of hearing threshold using this methodology.

We designed and optimized a protocol for untrained dogs, performed in a soundproof booth to minimize extraneous noise influence. A handler assessed dog behavioral response to sounds played in ascending intensity on a speaker, positioned orthogonal to the dog; an ethogram was used to define the spectrum of responses. Stimuli included 5 sounds perceived to be ecologically salient to dogs (dog barks, dog whimpers, bird whistle) and 3 warble-tone sounds used in human audiology (0.5, 4, 8kHz frequency). Stimuli were pseudorandomized within each category. The first group of dogs (group 1, n=13) underwent less standardized preparation/acclimation, whereby dogs were not tested at a standard time, and were not fasted. Group 2 dogs (n=12) were fasted, tested in the early morning, and had an expanded acclimation process.

Dogs responded more consistently to ecologically salient sounds, and response threshold within the spectrum of ecological sounds varied depending on frequency representation (dog bark response threshold was higher than sounds containing higher frequencies). Response thresholds to warble tone stimuli were higher than in normal humans, and responses were highly variable. Variability in responses was lower in dogs from group 2 compared with group 1.

Our optimized protocol for behavioral observation audiometry using ecologically salient stimuli results in consistent testing of response threshold in untrained dogs. Future applications could include the study of the impact of aging, ear pathology, or other disease on hearing thresholds in dogs.

1.0. Introduction

In humans, audiometry (a behavioral test of hearing) is standardized, and used as a diagnostic measure for hearing loss severity¹. In the conventional method of audiometry, the patient acknowledges perception of a signal (usually ecological broadband stimuli such as human speech, or pure tones at various frequencies), typically with a hand raise or button press. This method has been adapted for pediatric populations, that cannot perform adult responses to stimuli until approximately 5 years old². These adapted methods include Behavioral Observation Audiometry, and Visual Reinforcement Audiometry. Although less commonly used in current practice due to the widespread availability of auditory evoked potentials (AEP), Behavioral Observation Audiometry is the typical approach for preverbal children aged under 6 months. Stimuli are played through earphones or a loudspeaker and any physical response (startle response, head turn, facial expression change) that is time-locked to the stimulus presentation and is not observed in the absence of a stimulus is considered indicative of the sound being heard. As the child develops, more reliable operant-conditioning methods are employed. From 6 to 36 months of age, children can participate in Visual Reinforcement Audiometry, whereby a sound is played through a speaker or earphones, and a head turn (usually toward the direction of the sound) is taken as the response to the sound. The child is conditioned to task using visual rewards, often a moving toy or cartoon. Less engaging visual stimuli reorient the child’s attention between trials.

Hearing assessment in dogs has been reported using a number of different methods. AEP testing of hearing threshold is considered the “gold-standard” objective test for detection of hearing loss in healthy dogs (Shiu et al., 1997; Wilson and Mills, 2005; Scheifele and Clark, 2012; Hopper et al., 2024; Stanger et al., 2024), aged dogs with cognitive impairment (Fefer et al., 2022), and dogs with chronic otitis (Mason et al., 2013). However, AEP testing is relatively invasive, and not all dogs will tolerate testing without sedation or anesthetic agents, which may modify neurologic function, and pose health risks, particularly in older dogs or those with comorbidities.

Alternatively, behavioral methods of hearing evaluation have been described in dogs. Some studies have used operant conditioning to train dogs to respond when a sound is perceived, prior to completion of testing, typically using a looking behavior. Two published studies using this methodology successfully assessed threshold of response to white noise (Guerineau et al., 2022), and to tones at different frequencies (Guerineau et al., 2024). However, the extent of training required may be time consuming, and depending on the training capability of individual dogs, not all dogs may pass testing criteria. As one of our research directions is to study natural aging in companion dogs, the application of simple, untrained behavioral assessment assessing hearing threshold on a continuous or high-fidelity ordinal scale is a major goal. Only a small number of studies have studied hearing function testing in untrained dogs. One study used ecologically salient sounds (dog cry/whine, dog bark/bay, plate crash, siren), played at increasing volumes (in approximately 10 dB increments), to determine a lower fidelity hearing threshold using behavioral reactions of the dog as outcomes (Bognar et al., 2020). This study determined that dogs aged over 10 years had higher hearing threshold than younger dogs but did not describe specific responses thresholds to individual sound responses, nor describe the different sounds’ frequency distributions. Although not a hearing function test per se, another group studied the responsiveness of dogs to different human vocalizations, and noted variance in responsiveness in older dogs to vocalizations with positive or negative valance (Smit et al., 2019). These studies support the concept that hearing function in untrained dogs can be assessed using ecologically salient stimuli.

With this study we wanted to address the knowledge gaps that exist regarding best practices and implementation of behavioral observation audiometry methods to assess untrained dog threshold of response to ecologically salient sounds representing different frequencies, and to standard warble tones used in the assessment of frequency-specific thresholds in humans. The primary objective of this work was to develop and refine protocols to test hearing threshold to a variety of sounds in untrained dogs. A secondary objective was to determine if there were differences in threshold to sounds containing different frequencies.

2.0. Materials and Methods

2.1. Subjects/participants and ethical approval

Companion dogs (Canis lupus familiaris) of both sexes were recruited. An initial eligibility screening survey was disseminated via the University veterinary clinical studies coordinator to local dog owners. Exclusion criteria were body weight less than 10lb (due to size constraints of the dog ear protection), aged over 8 years (due to increased risk of presbycusis (age-related hearing loss) (Strain et al., 2016)), history of anxiety (related to separation or veterinary visits) that might impact behavioral outcomes, history of aggression towards people (that might affect animal handling), or if the dog commonly received anti-anxiety medications for veterinary visits or travel (that might modify behavior). Dog owners completed a validated dog hearing function questionnaire (Mason et al., 2013). Procedures were approved by the Institutional Animal Care and Use Committee, and our reporting adheres to current ARRIVE guidelines (Percie du Sert et al., 2020). Dog owners signed an informed consent form and provided a medication history at the time of enrollment. Dogs received no specific training prior to participation.

2.2. Procedures

Dogs had a binaural otoscopic evaluation performed by a veterinarian, and the better ear (based on inflammation, occlusion, visibility of tympanum) was chosen for testing; if no ear was deemed healthier, the tested ear was selected by coin toss. Dogs were photographed from the front and side in order to assign them to an ear position group based on published designations (Barber et al., 2020). During testing, the untested ear was covered using commercially available ear protection (EarPro, RexSpecs Mount Hope OH), modified to have only one ear covered; the other ear remained uncovered for testing.

Dogs were tested for threshold of behavioral responses to sound during one session in a double-walled soundproof booth that met the American National Standards Institute (ANSI) standard for maximum permissible ambient sound pressure levels³. The booth was located in the basement of a multi-use campus building in an urban location. Dogs were acclimated to the sound booth and ear protection, and acclimation was assessed using the Fear, Anxiety and Stress Scale (FAS Ladder, fearfreehappyhomes.com) prior to beginning testing. Dogs were allowed to begin testing if their score on the FAS scale was at or below level 1 (Mild/subtle signs). Group 1 dogs were not fasted the morning of testing, and the time of day of testing was not standardized. Group 1 dogs received 10 minutes of sound booth acclimation with the handler present. Group 2 dogs were all tested in the early morning and were fasted the morning of testing. A modified acclimation protocol was adopted whereby all group 2 dogs received 30 minutes of acclimation (10 minutes outside of the sound booth, 20 minutes inside the sound booth, both with the handler present). These adjustments were made based on exclusion of some dogs in group 1 for excessive sleepiness/inattentiveness during testing.

Dogs were seated orthogonal to a single speaker, 1 meter away. A handler manually restrained the dog gently by the chest and observed their behavior. The block of ecological sounds was presented first, with the warble tones presented second. Within each block, the order of the stimulus identity/frequency was randomized. Stimuli were presented in ascending intensity; the stimuli began at an inaudible level and were increased in level until behavioral responses were observed. A behavioral ethogram was constructed to define a behavioral response (Table 1), and behavioral observations were made in real time by the dog handler. The sound presenter was not visible to the dog. The handler wore sound attenuating earmuffs with a noise reduction rating of 30 dB to reduce the influence of handler perception bias on interpretation of dog behavior. Once a confident reaction was observed, 3 subsequent increments of sound were played to assess the dog’s reaction. Testing lasted approximately 15 minutes.

Table 1.

Ethogram used for defining a reaction to sounds.

Interaction Behavior	Description
Ear movement	Ear moving upwards, forwards, rotating, twitching
Vocalization	Murmur, whimper, bark, groan, or other type of vocalization
Head movement	Tilting head left or right, turning head towards speaker
Eye movement	Movement of one or both eyes towards speaker
Eyebrow movement	Raising eyebrows
Respiration change	Cessation of panting, cessation of close-mouth respirations or prolonged pause in respirations
Body position	Shifting of body toward speaker, sudden stiffening of body or increased alertness
Lunge	Attempt to evade gentle restraint of observer and make diagonal or forward motion toward speaker

Two types of auditory stimuli were used. The first group were ecologically salient stimuli. These stimuli were chosen as they were assumed to be salient to the dogs to encourage a response. Five sounds were selected: 2 dog barks, 2 dog whimpers, and a bird whistle. Recordings were obtained from an open-source platform (freesound.org) and edited into short clips using Audacity (sources and raw sound files are provided in Supplemental Material). Recordings were monoaural, .wav format, with a sampling rate of 44.1 kHz. The spectra of each sound are provided in Supplemental Material. All sounds were essentially broadband, but the dog barks had greater energy in lower frequencies, whereas the dog whimpers and bird whistles had more energy in higher frequencies; the bird whistle had the greatest high-frequency energy. With a sampling rate of 44.1 kHz, the highest frequency that can be encoded without introducing artifacts is 22.05 kHz. The speaker and amp system used for the ecologically salient stimuli (KEF Q350 speaker with Crown XLS 1002 amplifier) supported a flat frequency response up to 12 kHz. Due to the nature of the recorded files, there was a lower limit at which the sounds could be presented with good fidelity, approximately 30 dBA. To avoid floor effects, these stimuli were presented in a background of white noise. The white noise was presented at 45 dBA (measured at the approximate location of the dogs’ ear) for group 1 dogs and 35dBA for group 2 dogs. The reason for this change was to reduce the testing duration, as more stimuli had to be presented to dogs before a response was noted, potentially influencing fatigue in group 1 dogs. In addition, measurement of background noise in a variety of laboratory spaces (not the sound booth) identified that 35dB was closer to ambient background noise. The stimuli files were normalized such that the sounds became louder in sound pressure level in 3 dB increments (a doubling of sound intensity). Thus, the threshold measured was a signal-to-noise ratio threshold, a commonly measured metric in human testing that indicates the lowest level of a stimulus relative to the level of a background noise the person is able to perceive (Plomp and Mimpen, 1979; Killion et al., 2004; Billings et al., 2024). This measurement is calculated in dB SNR, by subtracting the noise level from the stimuli level.

The second group of stimuli were narrowband stimuli used in more standard audiometric testing. Because the testing took place in the soundfield, frequency-modulated (FM) tones were used, which vary (“warble”) in a narrow range around a center frequency. The center frequencies chosen for this study were 0.5, 4, and 8 kHz, representing the range of typical frequencies tested in humans⁴. These tones were played from a standard clinical Type I audiometer (GSI Audiostar Pro) and calibrated soundfield loudspeaker (RadioEar SP90, Middelfart, Denmark). These stimuli were presented in the standard 5 dB Hearing Level (HL) increments used in clinical audiometry). Because these tones are generated sinusoidal signals from an audiometer, the lower limit for presentation was defined as the lower limit of the audiometer (-10 dB hearing level; HL). HL is a normed scale used to measure human hearing, where the absolute sound pressure level at any frequency is corrected by subtracting a reference equivalent sound pressure level so that the average threshold at each frequency is 0 dB.

2.3. Data analysis

Results for a specific sound presentation for a dog were excluded if the 3 subsequent louder sound presentations failed to elicit at least one notable behavioral response, as this indicated that outcomes were highly inconsistent/unreliable. Data were analyzed in R and Graphpad Prism for Mac (version 10). Normality testing using the Shapiro-Wilk test was applied to data. To assess differences in threshold within the groups of ecological sounds or warble tones in all dogs, dogs in group 1 and group 2, we used a mixed effects analysis with the Geisser-Greenhouse correction, and Tukey’s multiple comparisons test with repeated measures for different sound types. To compare outcomes in group 1 versus group 2 and between ear position groups, we applied a two-way mixed-effects model (maximum likelihood method) with the Geisser-Greenhouse correction, and Sidak’s multiple comparisons test. To examine the effect of trial order on threshold for dog bark 2 and bird whistle, we used a one-way ANOVA.

3.0. Results

3.1. Participants

A total of 33 dogs were recruited. Eight dogs (from group 1) were subsequently excluded; the data from these dogs are not reported, leaving n=25 that contributed data for analysis. Of the excluded dogs, two dogs could not be completely tested because they either did not reach FAS endpoints despite acclimation (a 40-month-old male neutered Whippet), or because of unanticipated test environment availability constraints (an 84-month-old male neutered Brittany). One dog (a 41-month-old male neutered mixed breed dog) was inadvertently tested using incorrect sound stimuli. Five group 1 dogs were tested but either did not meet inclusion criteria (aged >8 years n = 1, a 104-month-old male neutered mixed breed dog, or presence of neurologic disease manifesting as a head tilt n = 1, a 61-month-old Labrador Retriever), or dogs were excessively sleepy/inattentive during testing (n = 3, a 35-month-old female spayed Standard Poodle, a 63-month-old male neutered Border Collie and an 80-month-old male neutered Anatolian Shepherd, 2/3 dogs were tested after 10am). These findings provided rationale for the adjustments to acclimation and time of testing implemented in group 2.

Summary demographic information for tested dogs (n = 25) is presented in Table 2. No dogs had received any sedatives or anxiolytic medications in the preceding 7 days before testing. One dog from group 2 was currently receiving systemic medications at the time of testing; a 28-month-old male neutered Nova Scotia Duck Tolling Retriever was receiving medications (omeprazole, ondansetron, pancreatic enzyme supplement, cisapride) for gastrointestinal disease; These medications were administered the morning of the testing.

Table 2.

Demographics for all dogs, and those in group 1 and 2.

	All dogs (n = 25)	Group 1 (n = 13)	Group 2 (n = 12)
Age, months – median (IQR)	43 (27–63)	35 (26–58)	47 (30–64)
Sex/neutering status	17/25 male (3 intact) 8/25 female (0 intact)	10/13 male (3 intact) 3/13 female (0 intact)	7/12 male (0 intact) 5/12 female (0 intact)
Bodyweight, kg – median (IQR)	22.5 (16.8–31.1)	22.9 (19.2–30.8)	21.3 (16.4–32.8)
Breed	12/25 purebred (3 Labrador Retriever, 2 Golden Retriever, 2 Shiba Inu, 1 each of Border Collie, Beagle, Great Dane, Siberian Husky, Nova Scotia Duck Tolling Retriever)	6/13 purebred (2 Labrador Retriever, 2 Golden Retriever, 1 each of Border Collie, Beagle)	6/12 purebred (2 Shiba Inu, 1 Labrador Retriever, 1 each of Great Dane, Siberian Husky, Nova Scotia Duck Tolling Retriever)
Tested ear	15/25 AS 10/25 AD	8/13 AS 5/13 AD	7/12 AS 5/12 AD
Hearing function questionnaire “score”( Hopper et al., 2024 ), median (IQR)	100 (87.5–100)	100 (87.5–100)	87.5 (87.5–100)

3.2. Hearing and ear assessment

The hearing function questionnaire responses indicated that the majority (24/25) did not meet the criteria for hearing impairment (score <76.1(Hopper et al., 2024), Table 2). For the one dog (a 26-month-old male neutered mixed breed dog, in group 1) whose owner questionnaire responses indicated hearing impairment (score 62.5), otoscopic examination findings were unremarkable, and threshold responses were not notably different from the other dogs in group 1 (data not shown).

Three dogs did not tolerate otoscopic evaluation – the tested ear for these dogs was based on external ear examination (selecting the ear with the least evident pinnal dermatitis; n=1) or coin toss (n = 2). Of the remaining 22 dogs, 1 tympanum was not visualized, 21/22 tympanums were deemed normal; 2/22 dogs had evidence of mild pinnal dermatitis; 1/22 had mild palpable ear canal thickening; 11/22 had waxy debris in the ear canal; 4/22 had mild inflammation in the ear canal; 4/22 dogs had mild ear canal occlusion (comprising less than 33% occlusion). No severe ear pathology was identified in any examined dog.

3.3. Behavioral assessment

For both group 1 and 2 combined, median FAS score at baseline was 0–1 (alert/excited/anxious or perked/interested/anxious) with an interquartile range of 0 (relaxed) to 0–1.

For group 1, 12/13 dogs were tested in the morning, of those dogs, 3/13 began testing before 10am, the remainder began testing between 10am and 12pm. Median FAS score at baseline was 0 (relaxed) with an interquartile range of 0 to 0–1 (alert/excited/anxious or perked/interested/anxious). At the end of the 10-minute acclimation period, FAS scores did not change in any group 1 dog.

For group 2 – 12/12 dogs were tested in the morning; all began testing before 10am. Median FAS score at baseline was 0–1 (alert/excited/anxious or perked/interested/anxious) with an interquartile range of 0–1 to 1 (mild/subtle). By the end of the 30-minute acclimation period 3/4 dogs with an FAS score of 1 had a reduction in FAS score; the median FAS score at 30-minutes was 0–1, with an interquartile range of 0 to 0–1.

Results of normality testing indicated that the majority of data were normally distributed (data not shown); parametric analyses were applied.

3.3.1. Ecological sounds

When all dogs were analyzed as a single group there was a significant difference in threshold between the 5 ecological sounds (mixed effects model, p<0.0001); multiple comparisons testing (Tukey’s) indicated that there was a significantly higher threshold to dog bark 1 and dog bark 2 compared with both dog whimper 2 and bird whistle (Figure 1A). For group 1 dogs, there were no significant differences in threshold between the 5 ecological sounds (mixed effects model, p = 0.054, Figure 1B). The variance in group 2 data was lower for all sounds than group 1 (Table 3). For group 2 dogs, there was a significant difference in threshold between the 5 ecological sounds (mixed effects model, p=0.0003); multiple comparisons testing indicated a significant higher threshold to dog bark 1 compared with dog whimper 2, and a higher threshold to dog bark 2 compared with dog whimper 2 and bird whistle (Figure 1C). There were differences in response thresholds between group 1 and group 2 dogs (mixed effects model, p=0.0003); multiple comparisons test indicated this was significant for dog bark 2, which had a higher threshold in group 2 dogs (p=0.04). There were no differences in response thresholds between dogs with “dropped” ear position (n = 12) versus those with either an “erect” or “semi-erect” ear position (n = 13, mixed effects model p=0.98, data not shown).

Figure 1: Ecological sounds stimulus thresholds (signal to noise ratio).

A: aggregate data for all tested dogs, B: data from dogs in group 1, C: data from dogs in group 2. Individual dogs are shown with dots and grey connecting lines, mean and standard deviation are shown in black for each stimulus type. * p<0.05, ** p<0.01 Mixed effects analysis with Tukey multiple comparisons test.

Table 3.

Summary data for all dogs, dB SNR, mean ± SD. The majority of outcomes were normally distributed. Post-threshold responses indicate the number of dogs (of all subjects) that the post-threshold louder sounds (3 in total) elicited a detectable behavioral response.

		Group 1	Group 2	All dogs	Number of positive post-threshold responses
					3	2	1	0
Ecologically salient stimuli	Dog bark 1	−1 ± 7.2	5.8 ± 4.7	2.4 ± 6.9	14	7	3	1
	Dog bark 2	−0.8 ± 9.0	8.3 ± 6.5	3.6 ±9.0	22	3	0	0
	Dog whimper 1	−2.2 ± 9.2	4.8 ± 4.5	1.6±7.7	14	6	2	3
	Dog whimper 2	−7.3 ± 9.9	0.75 ± 3.9	−3.3 ± 8.4	21	3	0	1
	Bird whistle	−8.3 ± 9.5	−0.25 ± 4.5	−4.4 ± 8.4	19	4	2	0
Warble tone	0.5 kHz	34.2 ± 16.1	35.8 ± 9.7	35 ± 13.0	22	1	1	1
	4 kHz	34.2 ± 18.7	32.1 ± 9.9	33.2 ± 14.9	16	6	3	0
	8 kHz	33.1 ± 16.8	35.0 ± 6.7	34 ± 12.8	19	4	2	0

For the two sounds (dog bark 2 and bird whistle) that had the widest range of thresholds measured (40dB) we examined if the order of presentation was related to threshold. There were no significant differences in threshold related to trial order for dog bark 2 (p = 0.95) or bird whistle (p = 0.19, one-way ANOVA; data not shown).

3.3.2. Warble tones

When all dogs were analyzed as a single group there were no significant differences in threshold between the 3 warble sounds (mixed effects model, p=0.90, Figure 2A). For group 1 dogs, there were no differences in threshold between the 3 warble sounds (mixed effects model, p = 0.96, Figure 2B). For group 2 dogs, there were no significant differences in threshold between the 3 warble sounds (mixed effects model, p=0.41, Figure 2C). The variance in group 2 data was lower for all sounds than group 1 (Table 3). There were no differences in response thresholds between group 1 and group 2 dogs (mixed effects model, p=0.85).

Figure 2:

Warble tone stimulus thresholds (dB HL based on human normalization data). A: aggregate data for all tested dogs, B: data from dogs in group 1, C: data from dogs in group 2. Individual dogs are shown with dots and grey connecting lines, mean and standard deviation are shown in black for each stimulus frequency.

4.0. Discussion

In this work we have developed and refined methods of behavioral observation audiometry testing in young, healthy, untrained companion dogs. Our studies identified that a combination of greater acclimation time, fasting for testing, and testing earlier in the day resulted in greater consistency of behavioral responses between animals. Thresholds for ecologically salient stimuli varied depending on the stimulus; stimuli containing higher frequencies had a lower threshold of response than those containing lower frequencies. Warble tone stimuli had high thresholds of response compared with humans, and response thresholds did not vary depending on stimulus frequency.

We found that the testing protocol affected test outcomes. In group 1, the timing of testing was inconsistent in relation to last meal and time of day, and we presumed because of this, the energy level and satiation of each dog also varied, which we hypothesize might affect attentional engagement with behavioral testing (Vanitallie, 2006). In addition, the extent of acclimation to the test environment in group 1 was relatively limited. The combination of these factors likely contributed to the inattentiveness of some of the dogs to the task. Even within group 1 dogs that completed testing, variance of thresholds between dogs was higher. After initial review of these data, additional measures were put in place to enhance dog attentiveness – dogs were fasted on the morning of testing, tested in the early morning, and were provided ample opportunity to acclimate to the ear protection and engage with the test environment prior to testing. Outcomes in group 2 had smaller variance, and no dogs were excluded due to inattentiveness. Trial order did not significantly influence outcomes for dog bark 2 and bird whistle, indicating that motivation during the testing window did not wane. Demographics of dogs in groups 1 and 2 were relatively similar although we cannot fully exclude the possibility the makeup of dogs in the different groups contributed to differences in variance. While these implemented protocol changes may explain the reduced variance, it is also possible that there was a learning effect on the part of the handler/observer over the transition from group 1 to 2, and/or the dogs chosen for group 2 were more homogenous in terms of behavioral responses. However, these findings underpin the necessity for consistency in testing, and consideration of the attentional state of the dog at the time of testing.

The standard deviation of SNR threshold values for ecologically salient stimuli were lower than the non-ecologically salient stimuli, supporting the potentially greater inherent salience of ecologically salient stimuli to dogs. Given what we know about the relationship between pure-tone thresholds and signal-to-noise (SNR) ratio thresholds for ecologically salient stimuli (e.g., speech) in humans (Smith et al., 2024), the SNR thresholds to ecologically salient stimuli in young dogs were generally typical of those in human listeners with essentially normal hearing (a SNR of around 0dB), but the pure-tone thresholds suggest (by human standards) moderate hearing loss (of on average >30dB HL). While we did not assess hearing function in an objective manner in these dogs, only one dog’s owner described behavior at home in the validated hearing function questionnaire that might suggest mild hearing loss. We predict that the lack of responsiveness to the warble tones was due to a lack of inherent salience to dogs and would therefore be less likely to induce appetitive or aversive instinctual behaviors (Beer, 2020). Similarly, in a study using electroencephalogram recordings, untrained dogs lacked detectable responses to heterospecific vocalizations (humans and pigs) compared with conspecific (dog) vocalizations (Morvai et al., 2025). Therefore, our findings and those of others suggest that in order to induce measurable behavioral response in dogs, sound stimuli must be perceived, and either appetitive or aversive.

In further support of their salience, we found a higher SNR threshold in ecologically salient stimuli that contained lower frequency sounds, versus ecological sounds with stimuli that contained higher frequency sounds which had a lower SNR threshold. This finding was not replicated in the different frequencies of the non-ecologically salient sounds. Using a trained two-alternative choice behavior method in trained dogs, two independent studies have shown that dogs have a lower threshold to higher frequency tone stimuli (16–20kHz) compared with lower frequencies (Heffner, 1983; Guerineau et al., 2024). Because dogs in these published studies were trained to express an overt, noninstinctual behavior once tones were perceived, we can assume that the lack of responsiveness in our study did not reflect a lack of perception, more so a lack of instinctual salience. Collectively, these dog behavioral studies might suggest that frequency-specific differences in thresholds reflect motivational-structural variance between low frequency sounds (produced more commonly in hostile contexts) and high frequency sounds (produced more commonly in friendly contexts)(Morton, 1977). Based on this theory, dogs appear more responsive to “friendly context”, higher frequency sounds. However, counter to this argument, a nonbehavioral study in dogs using brainstem auditory evoked response thresholds to tone stimuli found that hearing thresholds were also lower at higher tone frequencies (Ter Haar et al., 2008). Because we used open-source sounds, we were unable to determine the context in which the dog vocalizations were produced (for example in contexts of play, fear, separation, aggression). Others have shown that dogs recognize and behaviorally respond to differences in dog vocalization familiarity and context, including a familiar or strange dog reacting to a stranger at a fence or being left alone (Pongrácz et al., 2014). Different motivational states of dogs in aggressive, friendly or submissive contexts could result in acoustically different barks that carry context- and individual-specific information (Molnár et al., 2009). It is possible that differences in context of the different dog vocalizations (barks and whimpers) could explain the differences in behaviorally determined thresholds we detected. However, the bird whistle sound which had the lowest behavioral threshold would be anticipated to have less familiarity/contextual salience to dogs, supporting the hypothesis that at least some of the variance in response threshold could be attributed to aspects of stimulus frequency distribution.

Our ecological sounds were selected prior to the publication of a manuscript describing the dog soundscape (Savel and Legou, 2024). This study supports that the ecological sounds we selected are commonly found aspects of the dog soundscape, with bird song and dog barks ranking as some of the most common sounds in the dog soundscape. Dog whimpers and whines occurred less frequently in the sampled dogs’ soundscapes. These researchers also measured the frequency ranges for these ecological sounds and identified that dog barks had a relatively lower upper frequency (947 Hz) than dog whimpers (1357 Hz) and bird song (7773 Hz). This matches our findings that our presented dog barks represented lower frequencies than bird whistle and dog whimpers.

Our study had several limitations. We were underpowered to determine whether hearing threshold varied by sex, breed, or bodyweight. We did not retest dogs to determine test-retest reliability, nor did we compare outcomes with objective assessment of hearing using auditory evoked potential threshold response determination (Hopper et al., 2024). These are important future directions using outcomes for the ecologically salient stimuli and our modified test protocol for acclimation. Similarly, for future studies, a soundproof booth, while optimal, may be infeasible both in research and clinical settings, and determination of methods to facilitate testing in more typical clinical or research environments will be important. We did not test any dogs with substantive hearing loss, and an important future direction will be to determine whether outcomes of this test mirror age-related presbycusis previously identified using objective methods of hearing threshold determination (Ter Haar et al., 2008; Ter Haar et al., 2009; Strain et al., 2016; Hopper et al., 2024).