Checking the Claim: This Device Would Allow Dogs to Talk Like Humans | Innovation | Smithsonian
Current Issue
July / August 2014  magazine cover
Subscribe

Save 81% off the newsstand price!

EEG technology allows people to play music and control vehicles with their minds. But can it translate a dog's thoughts into words? (Nordic Society of Invention and Discovery)

Checking the Claim: This Device Would Allow Dogs to Talk Like Humans

A team of oddball inventors claims they are developing a headset that translates a canine's thoughts into words

smithsonian.com

In a way, the intimate relationship between man and man's best friend is unjustly lopsided. For their part, dogs are able to understand us very well. In fact, researchers believe a border collie named Chaser has demonstrated a vocabulary of more than 1,000 words, along with the ability to comprehend more complex language elements such as grammar and sentences. Meanwhile humans, despite even the most, er, dogged scientific efforts, have yet to decode the literal meaning behind a canine’s bark (if there is any).

But a Swedish design lab that calls itself the Nordic Society for Invention and Discovery thinks that animal behaviorists have been going about it the wrong way. What its developers are proposing instead is the development of a device that can infer what an animal is thinking or feeling by analyzing, in real-time, changes in the brain. The concept they've imagined, dubbed No More Woof, would be sold as a lightweight headset lined with electroencephalogram (EEG) sensors, which record brain wave activity.

When combined with a low-cost Raspberry Pi microcomputer, the inventors surmise that the electrode-filled device, which rests atop a dog's head, could match a wide range of signals to distinct thought patterns. A specialized software known as a brain-computer interface (BCI) would then translate the data into phrases to communicate. The phrases, played through a loudspeaker, may range from "I'm tired" to "I'm curious what that is." 

In December, the development team launched a crowdfunding campaign on Indiegogo.com in hopes of raising enough money to at least further explore the feasibility of such an idea (the BCI, for instance, is just an experiment at the moment). With a $65 donation, supporters of the project had an opportunity to reserve beta versions of the gadget, programmed to distinguish between two to three thought patterns, such as tiredness, hunger and curiosity, and communicate them in English. Those who pledged as much as $600 will receive a higher-end model capable of translating more than four distinct thoughts and suitable for a number of different breeds, which the group concedes has proven to be quite difficult. 

"The challenge is to make a device that fits different dogs and measures in the right place," says Per Cromwell,  the product's creator. "If it gets displaced it can lose the signal. We are struggling with these topics and would rather describe the devices we are working on as as working prototypes rather than mass produced products."

While developers more than doubled their initial goal—raising $22,664—you may not want to get your credit card out quite yet.

Since the Indiegogo launch, neuroimaging experts have come out to debunk claims made on the product's website, saying the science doesn't add up.

"What I saw in their video can't work," Bruce Luber, a Duke University professor who specializes in brain stimulation and neurophysiology, tells Popular Science.

Luber points out, for instance, that since EEG is designed to measure neural activity near the surface area of the brain, it won't be able to determine if an animal (or human) is feeling hungry; that feeling originates in the hypothalamus, which is located deep in the center of the brain. And while devices are being developed to allow users to move prosthetic limbs, steer a car or even play music, reliably identifying specific emotions and thoughts has thus far been beyond the scope of even the most sophisticated technology.

To be fair, Cromwell admits that the concept is being treated more or less as an experiment, or an exploration. There's also a disclaimer from the developers on Indiegogo that flatly states that No More Woof is still a work-in-progress and contributions do not guarantee a working product.

"When we started out we had no idea if it would work or not," he says in an email. "And to some extent we're still trying to make it work. So I think it would be more correct to describe the work as a couple of curious persons than being based on existing research.”

It's worth noting that this is the same oddball band of inventors to pursue other wacky ideas—from an indoor cloud to a flying lamp and a magic carpet for pets—but never deliver on them. Cromwell does claim to have made some progress, nonetheless, in pinpointing certain patterns he believes indicate, if not thoughts, at least a narrowed sense of what mood the dog is in.

The testing process, which he described in an email, involves using a video camera along with an EEG device to simultaneously record a dog's brain activity and physical response as it's exposed to a variety of stimuli, such as an unknown person, a ball, food or the smell of a treat.

“What we're focusing on in these early stages is measuring the amount of activity,” Cromwell explains. “Curiosity and agitation showed a significant increase in brain activity, and we're interpreting this as the dog being either curious and asking 'What is that?' or saying 'I want to play.' Conversely, when the dog is bored or tired, brain activity decreases and we translate this as 'Leave me alone' and 'I want to sleep.'"

Whether or not you find his method of translating dogspeak into intelligible words to be a stretch, Cromwell contends that it's an approach that should eventually lead to more accurate interpretation, as the team's research progresses. Currently, the only language option is English. “We know it's our translation and not an exact translation," he says. “But we are confident that more research will help us find and decipher more patterns.”

Will we ever see a machine that would allow human and pet to engage in actual conversation? If society wants it badly enough, it's totally possible, Luber tells Popular Science, particularly "if you get DARPA to put about $100 million toward it and get all of us working on it." 

Tags
About Tuan C. Nguyen
Tuan C. Nguyen

Tuan C. Nguyen is a Silicon Valley-based journalist specializing in technology, health, design and innovation. His work has appeared in ABCNews.com, NBCNews.com, FoxNews.com, CBS' SmartPlanet and LiveScience.

Read more from this author

Comment on this Story

comments powered by Disqus