Imagine you’re enjoying a pleasant stroll through the park when you hear someone say, “My owner needs your attention! Please follow me!” You look around and see no one; then you look down and see a dog staring up at you. You think someone’s pulling a fast one. But then the dog reaches around and his mouth tugs something on the yellow vest he’s wearing, and you hear it again.
“My owner needs your attention! Please follow me!” the dog seems to say again, looking at you plaintively and now beckoning with body language for you to follow him. You do, and he leads you to someone who’s having a severe allergic reaction, a seizure, or some other medical emergency.
Welcome to the new world of dog-human communication, where technology is allowing dogs to “speak” in ways we can easily understand when it really counts. If that same dog had run up to you Lassie-style and tried to get your attention without the voice, you might not have followed him. You might have reached down to pet him or thrown a pinecone for him while his person suffered alone nearby.
The vest is one of several emerging technologies a few universities are developing that could forever change the way working dogs (and probably pet dogs, down the road) can communicate with people—and the way people can communicate with dogs.
I wanted to see how these devices work, so I flew to Atlanta to visit the Georgia Institute of Technology (Georgia Tech), the epicenter of this field in the United States. One of the first researchers I met was Sky, a supersmart border collie who happens to be the dog of Melody Jackson, PhD, director of the school’s Animal-Computer Interaction Lab and also director of the FIDO project.* FIDO focuses on creating wearable technologies as well as other ways to open the lines of communication between dogs and humans. Sky is the star tester and demonstrator of the devices produced by Dr. Jackson and her team.
When Dr. Jackson and Sky greeted me in the lobby of the university’s genius-filled Technology Square Research Building, Sky inspected me with his intelligent border collie eyes and gave me a few sniffs. If he could talk (beyond the talking vest, that is), he might have said something like this:
Barely remembers long division. Why is she here? Does not compute! Don’t let her upstairs!
But he couldn’t. So we went up to Dr. Jackson’s office. Sky was wearing a bright yellow Georgia Tech vest and a matching Georgia Tech collar. He had the school spirit, but this was not the vest I’d come to see. When we got to Dr. Jackson’s office, she rustled up a couple of prototypes of the talking vest.
“Working dogs are smart, and they have important information they need to tell their handlers or someone else. But they’re limited in what they can do,” she said, handing me a small yellow vest with the words “FIDO research team” on the side. “This is just the beginning of helping them communicate.”
She explained that when the vest is ready for use in the real world, the electronics will all be covered and disguised. But for now, as it gets tweaked and improved, the guts are exposed. The talking vest, in this bare-bones prototype, would get a dog stopped at any security checkpoint. On the top of the vest is a plastic-and-metal controller box about the size of a deck of cards. Out of this stream red and black wires and a couple of white conductor ribbon cables. Some of the wires end in a hard plastic tube that protrudes from the side of the vest.
Dr. Jackson asked me to touch the blue surface of the tube. If I were a dog, I’d just tap it with my nose. I expected to hear the slightly robotic female voice I’d heard on a video of Sky using one of the vests.
“My owner needs your attention!” a man’s voice announced with a slight Southern twinge. “Please follow me!”
“A man? A Southern man?” I asked, laughing in surprise. Dr. Jackson explained that during testing, users said they wanted their dog’s voice to match their dog’s gender. So they got one of the department’s male researchers to be the voice of Sky’s technology. I envisioned the day the vests would come with customized accents and local parlance. Gus could have the California version, ending with “Dude, follow me!”
More important than the voice is the message. Dr. Jackson told me that in real life, the dog would be trained to “say” the emergency message twice.
“In our tests, usually the first time people hear it they jump back or just don’t believe what they’ve heard,” she said.
At this point the talking vest is a one-trick pony. But future vests will likely incorporate a few tabs and pulleys for dogs to “say” whatever information is most important for his or her job.
A seizure-alert dog could tell her person to find a safe place before a seizure by pulling on one tab. She could summon help if necessary by nosing another. This would be especially handy in busy public settings when the dog’s normal alert signal might not be as obvious to the handler and where there’s no life-alert button as there might be at home.
The messages could be customized depending on the disability. Diabetic-alert dogs who detect both low and high blood sugar might have an easier job of communicating if they can “say” their person is going high or low.
The technology could even apply to military working dogs and police dogs. After alerting to an explosive, Dr. Jackson said, the dog could communicate if it’s a particularly unstable type, like triacetone triperoxide (TATP), or something more stable, like C-4.
At the time of my visit there were no real-world applications of the technology in place, but that was about to change. The team was working on a vest for a service dog organization that wanted to try it for children with autism. A dog would be trained to use it to supplement a few tasks in his normal repertoire. If the dog sensed an impending meltdown, he could nuzzle into the child and use his usual body language to try to calm him or her. But with the talking vest, he could also gently ask the child directly, “Could you please pet me now?” or some such phrase. The vest will be a full-size service dog vest with the best sensors the team has developed.
The FIDO team has spent a few years creating the kinds of sensors that would stand up to dog mouths and a variety of environments, including water. Another prototype on Dr. Jackson’s desk is more friendly looking, with a simple tug toy attached to it. A paper describing the research on the tug-toy sensor in the Journal on Technology & Persons with Disabilities delves into the technical details of the model, then describes the tug-toy sensor as being equipped with a gloriously nontechnical-sounding Kong Wubba—a goofy name for a product sold by the Kong pet toy company. On the next page the paper gets technical again, with a description of voltage-divider schematics.
Other sensors developed by FIDO include one a dog has to gently bite to set off a voice, and another a dog just has to pass his nose next to, as we do with our hands under touchless faucets.
“We’ve come a long way, but we still have a way to go,” Dr. Jackson said as she recounted some of the frustrating and amusing issues they’ve faced.
There have been crunched sensors and drowned sensors. The proximity sensor was sometimes triggered by random objects the dogs would pass in their environment, or if they scratched an itch. These last two resulted in many false alerts, with dogs inadvertently calling for help as, say, they groomed themselves—the ultimate twenty-first-century version of crying wolf.
Georgia Tech scientists are no strangers to the challenges of creating reliable wearable technology. In fact, who should be sitting in the lounge area outside Dr. Jackson’s office but Thad Starner, PhD, founder and director of the school’s Contextual Computing Group. The wearable computing pioneer and technical lead for the infamous Google Glass was dressed in his trademark plain black T-shirt and black pants, eating a Subway sandwich and working on his computer, his smartphone, and a one-handed keyboard called the Twiddler—all at the same time, it appeared. He was also wearing Google Glass, the “smart glasses” he had helped develop.
Dr. Jackson and I sat down to talk with him briefly on our way to see Sky do a demo of some other communications prototypes. Dr. Starner has been instrumental in developing several of the FIDO devices. He has three advanced degrees from MIT. Dr. Jackson has two advanced degrees from Georgia Tech. It was heartening that these are the kinds of brains that want to help dogs better communicate important information to humans.
At the end of our conversation, Dr. Starner took my contact info. Or, rather, he found it somehow by looking into his Google Glass. Maybe an email had been cc’d to him? Then he asked me to look at him so he could take a photo with his Google Glass to go with my contact info. It was like magic—and slightly disconcerting—to have someone take a photo of me just by looking in my direction.
But that’s not what ended up happening. Google Glass didn’t take the photo. Something went wrong. Dr. Starner took off his glasses and tried to figure out the problem.
“Wearable computing is mind-blowing,” he said as he tinkered. “Especially when it works.”
I have little doubt this team will succeed in helping working dogs “talk” in a potentially lifesaving manner. But what about pet dogs? What if they could really talk instead of communicating with us the way they currently do with their eyes and body language and their own canine utterances? What if our pet dogs were able to “say” things they couldn’t before?
Isn’t part of the bliss of dogs that they listen to us as if they truly understand? We can tell them anything, and they don’t offer advice or try to fix things or blabber it to someone else. They’re just there, with their unconditional love and understanding.
Several years ago when I was news editor at Dogster.com, a Japanese company came out with a gadget that was purported to interpret barks. It was more of a gimmick than anything, but it started a conversation about whether we would really want our dogs to be able to talk.
Would Gus beg at the dinner table with more than his eyes? (Want! Please!) Would he suggest I stop using deodorant and toothpaste so I would smell better? (No, don’t do that! Stop!) Or would he, given the chance, just keep listening, only chiming in for important communications?
Taking this a step further, maybe in the future there will be vests for people who want their dogs to help them reach goals. I’d get the “writer on deadline” vest. Gus could be trained to sense by my body language or my scent that I was getting distracted and about to head to the fridge or checking my social media accounts. He would then pull the Kong Wubba on his vest. “Don’t even think of procrastinating! This book isn’t going to write itself,” it would say. Or an encouraging “You’ve got this!” And there’s always the popular “Sit! Stay!”
There wouldn’t be many phrases because the real estate on a talking vest is limited. There are only so many sensors that can be placed on a vest before it gets uncomfortable or unwieldy. So the FIDO team is working on other ways to expand what dogs can tell us.
The team designed motion-detecting collars, and they’ve trained dogs to “gesture” by moving their heads or bodies in certain ways. Mini electronics in the collar would interpret the gestures and put into words what the dog is trying to get across. It’s like the talking vest, only without a dog having to do anything with her mouth. It just requires a little choreography on the part of the dog.
And then there’s touchscreen technology. At this point the touchscreen the FIDO folks are using is the size of a medium flat-screen TV. The screen in a real-life scenario could be significantly smaller. Depending on its use, it might be only the size of a large tablet if portability is important.
Dr. Jackson and Sky showed me how it works. “Help!” Dr. Jackson said, undramatically, while looking at Sky. He immediately reached back to pull a sensor on his vest, but this wasn’t the right vest. Realizing he was outfitted in his nontalking vest, he performed his other “help” function without missing a beat.
He ran over to the screen, where the numbers “9-1-1” appeared in white, against three circles colored bright blue, yellow, and green. Sky touched them in sequence, but nothing happened. Dr. Jackson explained that his nose hadn’t hit them just right. He tried again, and this time the screen switched to a notification that an alert had been sent.
I asked if it would be easier to just have a large “HELP” button dogs could bump with their noses in an emergency. Dr. Jackson said it would be too easy, with dogs calling 9-1-1 or other emergency contacts if they accidentally touched it with their nose.
As service dogs like Terri Krake’s Brody can tell you, there are already devices dogs can use to summon help in an emergency. Brody, as you may recall, runs over and presses a big life-alert button with his paw if Terri is in trouble from a seizure. A notification goes to a chain of family and friends, or 9-1-1, or both.
The 9-1-1 feature of the Georgia Tech touchscreen is partly a trial to see if dogs can work with touchscreens and how touchscreens react to dogs. Slobber was a confounding factor for a while, but they’ve worked out some fixes.
FIDO researchers are investigating ways working dogs might be able to communicate using this technology. They’ve been exploring how dogs react to the colors, shapes, size, and placement of icons. The goal is to see if dogs can be trained to use a touchscreen to communicate information specific to their job.
Dr. Jackson says icons could be linked to a text or voice message for easy conveyance of what a dog is trying to communicate. She envisions touchscreen applications being a useful tool for dogs who help the hearing impared. They could alert their deaf handlers to different sounds by pressing icons that would send them text messages like “doorbell is ringing” or “tornado siren.”
Dogs have already proven to be rapid and enthusiastic learners of touchscreen technology. Researchers in Budapest and Vienna have trained more than two hundred pet dogs to nose-touch certain images on a screen. The results are promising, with researchers concluding: “The power of the touchscreen as a training tool is in its flexibility, reliability and controllability, and in its ability to provide novel motivational experiences. The number of cognitive training possibilities are limitless . . .”
Other researchers have used touchscreens to determine if dogs can discriminate between emotional expressions in human faces. With continuing research into dogs interacting with screens, it may not be long before your dog is asking (via her pleading eyes or a talking vest) for an iPad of her own.
Depending on what researchers learn in their ongoing quest to find the scent of cancers, dogs might be able to use icons or even an onscreen slider—like those we use for volume, only easier to manipulate—to communicate the strength of the scent. This might tell researchers something about the stage of the cancer. Or dogs could even identify the type of cancer if it turns out cancers do have distinct odors.
Applications go well beyond the health and service dog field. Bomb dogs might be able to give more specific information about what they’re smelling. They could tap their nose on an icon they’ve learned to associate with a specific explosive scent, and let their handler or the explosive ordnance disposal technician know what they’ve detected. Same with narcotics dogs.
It’s easy to see how touchscreen technology might evolve to become popular with people and their pets. Just the other evening, Gus bolted down to the front door and barked as if Godzilla himself (or herself? The debate rages on) were trying to get into the house. I was annoyed because I was delving into some complex research for the book and didn’t want to break my concentration. But he wouldn’t stop, so I headed downstairs to check what was going on. The stripe of fur along Gus’s spine had bristled to make him look vaguely larger, and he was barking and growling at something invisible to my eyes.
“What are you barking at?” I asked. In response he barked some more until I got him to stop and shooed him upstairs. This is where a Georgia Tech touchscreen of colorful icons could have come in handy.
I’d hand Gus his slobber-proof screen, where he could choose from the following options, symbolized by icons he’d been trained to recognize.
Cat!
Someone raiding our recycling bin!
Bad guy!
Raccoon!
Evil mail carrier!
Skunk!
Evil UPS worker!
Godzilla!
I have no idea!
In England, computer scientists are having a different conversation with dogs. It’s a conversation led by dogs, so instead of words, the dogs are conveying their side of the story with their noses.
The biodetection canines at Medical Detection Dogs seem to do an excellent job at detecting the scent of cancer, but founder Claire Guest felt there was a breakdown in the flow of communication that—if fixed—could make them even more accurate.
Claire met Clara Mancini, PhD, at an MDD open house several years ago. Dr. Mancini is the founder and head of the Animal-Computer Interaction (ACI) Lab and the Pervasive Interaction Lab at the Open University (UK). Her research focuses on animal-centered approaches to design and research. She wants working animals to be able to do their tasks in a manner that’s more natural to them. It’s better for the animals and could lead to improved results.
Claire told Clara that she and the other trainers felt they were missing opportunities for the biodetection dogs to be able to convey more information than just a simple yes or no. She thought if researchers could provide dogs with a new way of answering the question of whether cancer is present in a sample, the dogs might be able to tell them more.
They discussed the possibility of dogs being able to show not only whether a sample was positive or negative, but also how sure they were of their assessment. Some samples might be in between, and that’s important information to be able to convey.
Researchers think that in some cases an unsure response could mean a dog is picking up on very early cancer. If the trainers don’t know a sample has the VOCs of the beginnings of cancer, the dog won’t get rewarded and might be confused about what to do. MDD’s trainers were troubled by the times they could tell a dog had accidentally passed a positive cancer sample and seemed to want a do-over. A dog would flick her head in its direction or alert at the next position, which wouldn’t get her any reward. There are no do-overs, so data may not always reflect what a dog knows.
The ACI team tried to figure out the best way for dogs to communicate the nuances they might be detecting. They bounced around the idea of a keyboard system, where dogs could use their paws to tap a small number of large buttons to show whether a sample was positive, negative, or in between. They’d seen this kind of setup work for dogs to ask for food, water, or going out.
But this solution didn’t seem natural for the dogs. Dogs usually interact with objects using their mouths and noses, not their paws. The researchers wanted the dogs to stay focused to do their best job, and not come off a scent to perform an activity that could be distracting.
Eventually the researchers realized that the dogs were interacting with samples spontaneously, often pressing their noses more strongly into the ones that were positive and lingering longer. So they designed and developed an interface with pressure sensors connected to the metal plates that hold the samples. They recorded the degree of pressure the dogs were exerting with each sample and graphed the results. The beauty of it was that they gained more information without any additional training.
Instead of cancer samples, ACI scientists have recently been using amyl acetate, a compound that smells like bananas and is sometimes used as a flavoring agent. It’s better for initial research because it’s consistent, and easy to control the concentrations.
The results so far have been encouraging. It’s easy to see the positives, the negatives, and the ones where the dogs were unsure. Do-overs aren’t necessary with a system like this. Researchers can refer to the graphs to see how a dog reacted.
“I’m very happy with how it’s going,” Dr. Mancini says. “By engaging with animals on their own terms, our research can provide the opportunity to better understand animals, their capacities and perspective. . . . It is wonderful what you can learn from dogs if you just know how to listen.”
Dogs have been listening all along. It’s part of what makes their bedside manner so appealing.
People who work at the intersection of dogs and science believe that doctor dogs will soon be ready to take their place at a kind of global bedside; that dogs have what it takes to help keep the world safer from natural scourges—stopping the spread of deadly disease outbreaks before they become epidemics or, worse, pandemics. They may even be able to protect us from bioterrorism. It’s a matter of being able to train dogs, test them, and deploy them where they’re needed.
“Dogs are portable rapid detectors. There’s real potential for expanding their detection of biological targets,” says Dr. Angle, of Auburn University. “But there are real challenges, too.”
In a paper in Frontiers in Veterinary Science, Dr. Angle and his coauthors urged researchers to realize “there are intricacies in training dogs on biological targets that are complex and unlike any skill set utilized in other aspects of dog training. A detection dog is a highly complex sensory technology, and not understanding its full capabilities and its influences could skew study results to not reflect the true potential of a dog.”
It’s up to us to get it right. Fortunately, dogs, being dogs, will help us along the way.
What would it look like to have dogs on the front lines of containing a disease?
No one can say for sure, but it might be something as simple as dogs working at a checkpoint for malaria and sniffing everyone who passes through. Think airport dogs, only in a variety of settings, and sniffing for disease, not explosives. If an explosives-detection dog at an airport alerts on someone, that person will be thoroughly checked out before being permitted to proceed. Similarly, if a malaria dog alerts on someone, the person would probably need to be tested before being allowed to enter an area that has eradicated or greatly reduced malaria.
It gets more complicated for some other diseases. Take the Ebola virus. Instead of checkpoints where dogs would smell people passing through, Dr. Angle thinks dogs could expedite testing of people behind the scenes.*
If dogs could be trained to detect Ebola from cheek swabs placed inside individual containers, for instance, they could sniff large quantities quickly. As with malaria, if a dog alerts, the donor would need further testing.
“They can’t run blood tests on a thousand people moving through an area,” Dr. Angle says, “but they can do a quick swab, and dogs could act as an initial screen. It could greatly help containment.”
It’s going to take time and a deep commitment and investment in research to get to the point where dogs could one day help save lives on a massive scale. Fortunately, the work has already begun.
The pathogen hunt has started in hospitals, but walls won’t contain these doctor dogs for much longer. If all goes well with the malaria study, we could be seeing dogs helping beat down this disease in Africa within a few years.
As dogs succeed, there will likely be no shortage of ideas for how they can help detect other diseases. And not just ones that cause illness in humans. Dr. Angle thinks dogs could be invaluable in detecting pathogens in livestock, crops, and soil.
Meanwhile cancer-detection dogs will be hard at work in their quest to find the odor of cancer in exchange for their favorite toy or treat. Researchers will keep looking for ways to improve screening. More researchers from a variety of disciplines are joining the quest all the time.
I recently received an email from the Royal Dutch Guide Dog Foundation, which provides guide dogs and other service dogs and trains cancer-detection dogs. I had visited the Amsterdam-area organization and watched its dogs sniff for colorectal cancer as part of a pilot study. The group has now been asked by the Leiden University Medical Center to see if biodetection dogs can find cancer in blood samples of women under age forty who have a common breast cancer gene. Patrick Hilverink, the head of medical detection dogs for the foundation, wrote me that mammography is “far from effective for this group, with only 40 percent sensitivity.”
Their goal is to find breast cancer early and reliably. “It would of course be fantastic if these dogs could make a valuable contribution here and save lives,” he wrote.
If all goes well, those of us at higher risk for cancers that are hard to detect early will be able to have our blood, urine, or saliva analyzed by a reliable dog scan that will give us peace of mind for six months or a year or more. (This is assuming that researchers can get past the significant challenges of screening discussed in Chapter 3.) If something like this were around for ovarian cancer, I’d be less likely to consider going under the knife prophylactically.
Dr. Angle understands this line of reasoning: “If someone is on the fence 50-50 about having an organ or gland removed because of their cancer risk, and there were solid data showing dogs are 85 to 90 percent capable of detecting it at an early stage during screenings, they might opt to keep getting checked by cancer-detection dogs.
“In your case,” he told me, “you might end up being able to kick the can down the road if dogs give you the clear.”
We’re not there yet, though. Right now, because of the lack of good testing for this cancer, I’m kicking the can in the dark. I could end up over a cliff if I’m not careful. I’m not alone. There are people with much higher cancer risks than mine who would benefit even more than I could from having dogs detect cancer early.
In the war on cancer, dogs could be there to keep as many people as possible from danger—just as they do on the battlefields. It’s a comforting and hopeful thought.
But with the way e-nose technologies are advancing, cancer-sniffing devices may be here before the dogs deploy as a significant force against cancer. The olfaction technologies dogs are helping develop could have far-reaching impacts—even beyond cancer.
MIT’s Dr. Mershin thinks the e-nose his team is working on for cancer could easily be adapted for other uses.
“There’s no limit. Remember that if it smells, we can find it. Whatever dogs’ or other noses have ever been used for before, nanonoses can be used, only they will be faster, better, and eventually cheaper—although they are currently prohibitively expensive, same as early cameras and computers.”
He’s in good company with his ideas about expanding e-nose technology beyond cancer. A certain queen of the United Kingdom recently brought this up to him.
On the day Medical Detection Dogs gave its demonstrations at Buckingham Palace’s Royal Mews, Dr. Mershin was among the lucky attendees who got to meet Queen Elizabeth.
“I was blown away,” he says of their meeting. It wasn’t necessarily her refinement and grace that won his admiration. It was the way the nonagenarian monarch seemed truly and intelligently interested in the miniaturized version of MIT’s e-nose.
“She held the nose in her hand. Her elegant, gloved hand. And she asked me really good questions. She was genuinely curious. She wanted to know how the dogs are working to help what we’re doing.
“She even asked if we could put a device like this in airports for malaria,” he says. “I’ve met a lot of celebrities at MIT. They’re great, but you don’t expect to get ideas from them. But here was the ninety-two-year-old queen with this brilliant thought.”
His answer was yes, absolutely. This delighted the queen.
It seems we all love a good dog story. But if the happy ending involves small, lifesaving devices that dogs helped humans invent, that’s OK, too.
One day we may look back at this era with nostalgia. We will tell our children or grandchildren, who may tell their children and grandchildren, that there was once a time when people were getting sick and dying from many diseases, but that our best friends were at our side, helping us figure out what was wrong. The dogs did what they loved to do—they used their incredible noses—and we rewarded them with love and toys and treats.
We communicated as best as two species could, and eventually the dogs led us to the scents of danger. They helped us develop advanced machines that became widely available and easy for almost anyone to use.
Thanks to the dogs, and the researchers who believed in them, the world became a healthier place.