I am long overdue to write about this. I successfully desensitized and counterconditioned my clinically sound phobic dog, Zani, to electronic beeps. Here are some concepts and practices that could be helpful to others who are working with such dogs.
It’s Not Always about Volume
If I could convey one thing to people who want to desensitize their dogs to sounds other than low-pitched booms and bangs, it would be this: Think beyond the volume control.
We assume that the way to make a sound less intense to start desensitization is to turn down the volume. That makes sense for sounds where it’s the volume (and suddenness, usually) that make them startling.
For example, it’s probable that volume is relevant for thunder and fireworks. They are loud and sudden enough to trigger the mammalian acoustic startle response. And the startle response can trigger fear conditioning (Götz & Janik, 2011). It’s a fair starting point to assume the loudness and the suddenness are integral to a thunder-phobic dog’s response.
Now, what about that low-battery chirp of a smoke alarm that terrifies some dogs or the digital beep of a bathroom scale? Do we really think making these sounds quieter will make them less scary for phobic dogs? They’re not loud to begin with.
What Makes a Sound Intense?
To make a less intense version of a sound, we need to consider what might make it “intense” to a dog. We can’t know for sure, but science can help us make an educated guess.
Let’s consider the characteristics of a quiet digital beep and why it might scare a dog. We can start by looking at its waveform.
This is the beep of a bathroom scale.
This waveform image shows a beep about 0.15 seconds long. The x axis is time, and the y axis is amplitude. There are some striking things about this sound.
It is sudden. There is no gradual transition between off and on. It starts instantly. It’s homogeneous until the short fade at the end.
For comparison, the following is the waveform image of a bird chirp of about the same frequency.
The bird chirp is longer in duration, but the striking thing is how much more complex it is. And even though a chirp is a sudden noise too, you can see the gradual attack (audio term for the beginning of a sound). It’s different from the sudden start of the digital beep.
Back to the beep. There are a couple of other things we can learn about this sound through analysis. We can learn its frequency. It’s about 3,900 Hz; that is not visible on this kind of diagram. And since it is being generated through standard consumer circuitry and will play through a consumer speaker, it will not contain any tones higher than 20,000 Hz. That means the sound may sound odd and truncated to animals like dogs who can hear up to 40,000 Hz.
How does this add up? We don’t know why some dogs fear sounds in a particular frequency range. But we can make some conjecture about why some of the other features of this sound could add up to “scary.” According to research, dogs don’t locate sounds as well as we do (Fay and Wilber, 1989, p. 519). Add to that the short duration of the sound; shortness makes sounds harder to locate for everybody. Plus, these beeps are often pure tones, and that can be a challenge, too. Per Barber et al., “In general, it is easier to detect broadband sounds than pure tones.” This means that pure tones and those with high frequencies omitted could be harder to locate. Finally, “…so it may be possible [for a dog] to estimate the distance of a sound source only if the sound source has an expected volume” (Barber et al., 2020). In other words, they can locate it better if it’s not too quiet.
To be clear: the above conclusions are conjecture. They are based on some known information, but the conjecture doesn’t have experimental evidence yet to support it.
Have you ever searched for the smoke alarm emitting the low battery chirp when you have several smoke alarms? It can be maddeningly difficult. Now imagine if, like a dog, you had less skill at locating sounds. And the sound was weird and missing a lot of frequencies that would be present in an analog or natural sound. Not to mention that every time it happened, it was finished practically before you realized it had started.
To dogs, these sounds are likely hard to locate. Making them quieter could make the situation worse, not better. I have observed that to be so with my own dog.
How Can We Reduce the Intensity of a Beep?
The problems with volume sound like bad news at first. Adjusting volume is the easiest way to change a sound; we just turn a knob or drag a slider. But the good news is that there are lots of ways we can change a digital sound to find a way to make it less scary.
Here are some examples.
All the following short audio files play a “pure” sinusoidal beep first, then the altered beep. Make sure any beep-sensitive dogs are not anywhere near when you play them, even if you are wearing earbuds or headphones.
We can do any of the following, alone or in combination.
Change the frequency. In the case of a beep, it will usually mean lowering it.
Change the duration. In the case of a short beep, it will usually mean making it longer. That’s counterintuitive, but in keeping with the location challenges I’ve listed above.
Make it less sudden. It’s perfectly possible to alter sounds so they have a more gradual onset.
Make it less “pure.” That means to add frequencies or change the timbre some other way. You could add frequencies digitally, or use a more natural sound, say, a recording of a flute in the same range as the beep. One dog I helped with couldn’t tolerate a flute, but was OK with an oboe sound. For this recording I used a recording of a note on my harpsichord, altered to raise the pitch a bit. (It’s a lower frequency than the other sounds, to make it easier for our human ears to tell the difference between the digital beep and the harpsichord note.)
Mask it. “Hide” the sound in a white noise mask, and gradually remove the mask in the successive recordings. In this recording, I left the beep audible under the mask, but it can be started at an inaudible level. I wouldn’t use this method for a beep anyway, but masking is great for broadband noises like engines or even door slams.
Filters can be great tools, as well. There are many more kinds of sound edits we can do, singly or combined. Check out this screen shot of some of the options in the Audacity sound freeware. Not all work for our purposes, but many can.
Getting Back to the Original Sound
So we found a starter sound that doesn’t scare our dog. We can condition the dog that it predicts great things. What then? That’s not the sound they were afraid of. But we are working with digital sounds, so it’s just a math problem to get back to the original. We change the sound in gradual approximations back to the original sound. That’s the analog of starting quietly and raising the volume. And if we make more than one category of change to the sound, it may take more alterations to get back to the original sound.
I use Audacity to edit sounds. Having a musical background is great for this, but I think anyone who can discriminate pitch and timbre and who is comfortable with technology could learn to make a series of sounds in this way.
High Fidelity Digital Sounds
If the dog is afraid of a digital sound, as opposed to a sound in nature, there is an advantage to that. We can replicate such a sound very well on digital equipment. I mentioned above that speaker outputs cut off at 20,000 Hz. There is no reason for our human-oriented speakers to play anything higher. (Some audiophiles would argue, but that’s not a relevant discussion.) All sounds, digital and otherwise, rendered on consumer equipment will have those frequencies missing.
Sounds in nature include those higher frequencies (and super-low ones), so we can never replicate them perfectly by playing them through a speaker. But we can replicate digital sounds very well, even for dogs. If a dog fears a sound from a smart phone, we can record the sound and we can play it (and its derivatives) back on the smart phone. Being able to replicate the sound accurately gives a huge advantage over, for instance, trying to condition a dog to the sound of thunder using speakers.
I came up with these ideas independently and I’m not aware of anyone else in the dog world doing sound conditioning in this way. But the method is squarely within what we already know about behavior science and bioacoustics; it isn’t “New and Different.”
After I started implementing the method, I discovered there are a couple of research papers that describe success desensitizing to sounds using a variable other than volume. One was by Poppen (1970). In this experiment, rats were taught to associate a 3700 Hz tone with electric shock. Then they were exposed to a much lower tone (400 Hz) not coupled with shock, which was raised in five increments back to 3700 Hz. Some of the rats had the desensitization exposures alone, and some were also counterconditioned with food. Both groups “unlearned” their behavioral fear response, with the rats that received counterconditioning doing so faster. (This experiment used conditioned suppression, which I’m not going to explain here. But that’s how the scientists were able to measure the acquisition and extinction of fear.)
So it’s been done by scientists. I’ve done it, too. Zani was diagnosed with clinical sound phobia and was under the care of a veterinary behaviorist. I did the conditioning after she was stable on meds. I’ve embedded Zani’s “before and after” video here. Then in Part 2, I’ll present a mini-case study describing what I did, including a list of the sounds I used and a video showing many of the sound exposures.
Note: I’ll discuss this more in the next post, but I am no longer accepting clients for this work. But I want people to know that it can be done, with caution and under controlled conditions. And I plan to provide more resources.
Barber, A. L., Wilkinson, A., Montealegre-Z, F., Ratcliffe, V. F., Guo, K., & Mills, D. S. (2020). A comparison of hearing and auditory functioning between dogs and humans. Comparative Cognition & Behavior Reviews, 15, 45-94.
Fay, R. R., & Wilber, L. A. (1989). Hearing in vertebrates: a psychophysics databook. Hill-Fay Associates.
Götz, T., & Janik, V. M. (2011). Repeated elicitation of the acoustic startle reflex leads to sensitisation in subsequent avoidance behaviour and induces fear conditioning. BMC neuroscience, 12(1), 1-13.
Poppen, R. (1970). Counterconditioning of Conditioned Suppression in Rats. Psychological Reports, 27(2), 659–671.
Copyright 2023 Eileen Anderson