Searching for the Uncanny Valley

août 27, 2019

Your eyes are smoldering
As they look up at me devilishly;
Don’t be so full of accusations,
After all, a delicacy is a delicacy,
And your eyes, well, coldly gelled
They’ll go down so nicely,
If only you’d just stop staring.

Sue Flay

They used to joke that he should be doing more important things with his time.

When Masahiro Mori started teaching at the Tokyo Institute of Technology, Japan wasn’t very interested in robotics. In fact, no one would even dare making a grant proposal to the Ministry of Education to start a project on robots. As he explains in an interview with IEEE Spectrum, an applied sciences and engineering magazine, it was “too embarrassing.”

“They thought that it was frivolous to be working on a toy.”

No one seems to think that anymore. Masahiro Mori went on to become a pioneer in robotics. He’s the founder of Robocon, the first robot-building competition in Japan, and the director of Mukta Research Institute, studying the relationship between robots, spirituality, and religion.

At 92 years old, he doesn’t build robots anymore. But he still offers consultation to corporations on robotization, and he’ll still enthusiastically show all kinds of robot designs in his old notebooks. “Let me show you a sketchbook that I drew back in those days. These are from the 1960s before Xerox photocopies became available in Japan.”

Building eerie robots

Ever since he was a kid, Mori hated looking at wax figures. There was something that creeped him out, although he couldn’t exactly pinpoint what, or why. But it was his fascination with strange things that led him straight to the Uncanny Valley.

Almost 50 years ago, in 1970, Mori, then a 43-year-old researcher, published an essay in the obscure Japanese journal Energy, called Bukimi No Tani Genshō. In it, he predicted that as robots become more human-looking, they become more likeable. That is, up until a point.

Energy periodical cover, 1970

Take industrial robots, for example. As Mori writes in the original article, their design is clearly based on functionality.“These robots just extend, contract, and rotate their arms; without faces or legs, they do not look very human.” And therefore, we don’t feel much affinity for them. But as you add more human-like traits — two arms, two legs, a torso, and a friendly, wide-eyed expression— we start enjoying them more. That’s why we love Wall-E and his vulnerable, hopeful gaze, or Baymax, the inflatable nurturing robot in Big Hero 6.

But then, as we come closer to human-like appearance, you inevitably reach the Uncanny Valley — an uncomfortable middle ground where the grass ain’t green and the robots ain’t pretty. We can always tell when a robot has crossed the line — even when they look almost indistinguishable from real humans. There’s just something not quite right with them. Perhaps it’s a slightly off smile. A sudden twitch in the head. Perhaps it’s just the eyes, that never seem to meet yours. Whatever the culprit, there is always a precipitous drop in likability. Suddenly your skin is crawling with discomfort.

“One might say that the prosthetic hand has achieved a degree of resemblance to the human form, perhaps on a par with false teeth. However, when we realize the hand, which at first site looked real, is in fact artificial, we experience an eerie sensation,” Mori writes.

The Uncanny Valley didn’t draw much attention when it was first published. But in 2005, it was picked up at IEEE’s Conference on Humanoid Robots, and since then, it has been studied by a hoard of roboticists, psychologists, philosophers, anthropologists, designers.

And although some dismiss it as too simplistic, psychologists at Stanford and the University of California have found reason to believe it’s true. They studied the reactions of workers at a crowdsourcing platform to 80 real-world robot faces, who had to rate them on how mechanical or human they looked, and how pleasurable or enjoyable they thought it would be to interact with them every day. By plotting these ratings on a chart, the researchers found the Uncanny Valley — as robots become more lifelike, their perceived friendliness rises, then falls significantly, then rises again.

“Navigating a social world with robot partners: A quantitative cartography of the Uncanny Valley,” a research paper by Maya B. Mathur and David B. Reichling

At the end of his article, Mori wonders if there’s a reason for this eerie sensation we’re equipped with. “Is it essential for human beings?”

On the nature of creepiness

This feeling is not exactly easy to explain, but we’re all familiar with it. And that’s why Francis T. McAndrew, a social psychologist and professor at Knox College, decided to study it.

“I just started asking people, well, when you use that word, what do you mean? Does it mean you’re afraid?” And people would tell him that well, no, it’s not quite the same as being afraid. He would then ask “Does it mean you’re disgusted?” But no, it wasn’t quite that either. And yet, he tells me, “there seemed to be an agreement that there is this thing called creepiness.” But when he decided to look at what other psychologists and researchers had written on the subject, he couldn’t find anything.

“There wasn’t a single study on it.”

So he decided to study it himself. With the help of a student, Sara S. Koehnke, he had 1341 participants, with ages ranging from 18 to 77, rating the perceived creepiness of different behaviors and physical characteristics, such as “greasy hair”, or “never looked [my friend] in the eye.” They also asked participants to rate the creepiness of different occupations, hobbies, and in the fourth and final section, asked them to agree or disagree with statements to uncover why they felt that way, such as “I am uncomfortable because I cannot predict how he or she will behave.”

Although their evidence is completely anecdotal, McAndrews warns, the results are consistent with the hypothesis that being “creeped out” is an evolved adaptive emotional response, a heightened state of anxiety, to the ambiguity about the presence of threat. “A by-product of evolved human psychology,” he adds, “that enables us to maintain vigilance during times of uncertainty.”

Uncertainty being key here. If someone points a gun to your head, you’re clearly in danger, terrified — and rightly so —, but you would hardly describe the situation as creepy. “It’s not the clear presence of danger that makes us feel creepy,” he explains. “But the uncertainty of whether danger is present or not.”

He found other interesting things. For example, according to its 1341 participants, and something most of us would agree with, the number one creepiest occupation is clowns, followed by taxidermists and sex shop owners. Clowns have been in the spotlight lately, following the recent adaptation of Stephen King’s It, and the 2016 clown sightings in the US, but there’s a reason clowns started appearing in horror movies in the first place.

“They’ve got all of the things going on that would put us on our guard,” McAndrew tells me. “You can’t really tell what their emotions are. They have this painted smile and exaggerated features, the hair, the nose, the shoes. And they’re designed to be mischievous. If you go to a circus and a clown pulls someone out of the audience, you know nothing good is going to happen.”

They’re unpredictable. They don’t play by the rules. “And if they don’t understand those rules, what other rules might they break?”

I’d rather not dwell on it. But when I watch an awkward Sophia greet the audience during Web Summit to talk about Artificial Intelligence, I don’t feel quite the same dread as when Pennywise the Dancing Clown creeps in the drain, gnawing on the arm of six-year-old Georgie and dragging him to the sewers, not to be seen again. The Uncanny Valley can’t quite be explained just by the threat ambiguity. When McAndrew first decided to study the nature of creepiness, he didn’t think about the Uncanny Valley. But he has since given it a lot of thought.

“It’s slightly different,” he explains. “It’s a different kind of uncertainty. It’s still about not being sure what’s going on, but you’re encountering something that acts as a human being, and yet, at the same time, you consciously know that it’s not one. And so you’ve got these conflicting thoughts going on that create a tension. We’re uncomfortable because we can’t categorize it.”

Beyond categorization

One of these things is not like the others
One of these things just doesn’t belong
Can you tell which thing is not like the others
By the time I finish my song?

German Psychologist Jentsch also believed we feel uneasy about things that defy categorization. In his 1906 essay ‘Zur Psychologie des Unheimlichen’, translated to ‘On the Psychology of the Uncanny’ for the first time in 1995, he explains what has since been named the Categorical Uncertainty Hypothesis. Jentsch claims he’s not trying to define the essence of the uncanny, but rather find out why we feel that way. The only definition we’re offered is of the word “Unheimlich” itself.

“With the word “unheimlich”, the German language seems to have produced a rather fortunate formation. Without a doubt, this word appears to express that someone to whom something uncanny happens is not quite at home or at ease in the situation concerned, that the thing is or at least seems to be foreign to him.”

Familiar things are dear to us. There’s comfort in tradition, in routine. However, Jentsch notes that we deal with the unusual with “mistrust, unease and even hostility.” As long as the doubt as to the nature of the object lasts,” he writes, “a feeling of terror persists in the person concerned.“

Despite the fact that this essay was written over a century ago, Jentsch cleverly points out the design flaw that would be discussed in Mori’s essay 60 years later, and ever since: “It is of considerable interest to see how true art, in wise moderation, avoids the absolute and complete imitation of natural and living beings, well knowing that such an imitation can easily produce uneasiness.”

A perversion of humanity

When this eeriness is desirable, one of the most reliable devices for producing it is to leave people uncertain about the nature of the character standing in front of them. Is it a human? Something else entirely?

Some researchers suggest that the reason behind the Uncanny Valley is a violation of the expectation of humanity. An abomination in the natural order of things. As Mori described with the prosthetic hand, there’s a mismatch between the humanism of the hand, and its non-human behavior — the coldness, the texture, the limp boneless grip.

Some researchers have gone further with the Mortality Salience hypothesis. They argue that these human-like replicas remind us of our own mortality, “eliciting the uncanny feeling driven by a fear of death.” And it makes sense, especially when you think about cadavers and zombies.

Given that the Uncanny Valley elicits such disturbing feelings, why would anyone risk it?

To this day, Masahiro Mori still believes the Uncanny Valley shouldn’t be crossed. It’s not that it’s not possible, it’s just that, to him, it’s not worth it. “It’s possible to create affinity by deliberately pursuing nonhuman designs,” he tells IEEE. Take Asimo, for example, a 100 pound, 4 ft 3 inches tall humanoid designed by Honda in the turn of the century by one of his own students. Despite some clearly anthropomorphic traits — two legs, two arms, a torso, a head — and its countless updates over the years, Asimo is not necessarily trying to look human.

“I always tell them to stop there. Why do you have to take the risk and try to get closer to the other side? It’s not even interesting to develop a robot that looks exactly like human, from my perspective,” he adds.

With advances in AI, Virtual Reality, and even game design, we really have to ask ourselves whether the Uncanny Valley is worth crossing. Maybe our God complex isn’t worth an experience that leave people’s skin crawling.

Whether we decide to keep pursuing human-like robots in the future, or follow Mori’s advice and go forth a more creative route, if there’s one thing the Uncanny Valley tells us, is that we don’t like uncertainty.

“We like predictability,” McAndrew says. “And anything that makes it difficult for us to categorize things, or to know whether we’re facing a threat, is unpleasant and it motivates us to do whatever we can to clear it up. So I think one thing that tells us is we like clearer answers. “

And if we can get them without hellish robots creeping in our dreams (and lives), the better.

The post Searching for the Uncanny Valley appeared first on Unbabel.

About the Author

Profile Photo of Content Team
Content Team

Unbabel’s Content Team is responsible for showcasing Unbabel’s continuous growth and incredible pool of in-house experts. It delivers Unbabel’s unique brand across channels and produces accessible, compelling content on translation, localization, language, tech, CS, marketing, and more.

DeutschFrançaisNLdanskSvenskaEnglish