Uncanny Valley Explained ｜ Why Robots, Dolls and Mannequins are Creepy

Oh, so the wife and I are watching Christmas Chronicles with Kurt Russell as a singing Santa, right? We're having a good time. Then the movie introduced something truly terrifying elves. That was it. Movie night was over. I couldn't watch anymore. I was done. Oh, here we go. What's your problem with elves? Gimli? Well, the problem is that the elves looked almost real, but not quite almost natural, but not quite. And it was upsetting. I mean, haven't you ever seen an animation and felt a little bit creeped out? I'm looking at you, Polar Express. Well, or have you ever seen a mannequin or a robot or even a wax dummy that was very lifelike but still looked wrong? Sounds like you're describing a Silicon Valley CEO. Robotic and lifelike, but not human. Yep. To the point where it's uncomfortable even looking at them, much less being around them. Uh huh. Yep. They're awkward to watch. I think we've been going now for just under two hours. This anxiety is called Uncanny Valley and everyone experiences it, though some of us are less tolerant than others. Oh, let me guess. You are not very tolerant. You are correct. I am very susceptible to Uncanny Valley and I'm not alone. Let's find out why. Welcome to the Y Files, where smart folks like us come to laugh and learn. Today we're going to talk about Uncanny Valley and why we feel so anxious when we experience it. So let's start with the uncanny part. Uncanny is used to describe something that's almost identical or almost perfect, but not quite there. And that's what's happening with Uncanny Valley. Think of a movie you've seen where the CGI was almost human, but not quite. How did that make you feel? A bit uneasy, right? I remember Polar Express. This movie has become a legendary bomb. And most people agree it's because the animation style landed right in the middle of Uncanny Valley. The problem comes when something artificial like a robot animation statue or even a painting depicts the human form. Realistically, we're OK up to a certain point of realism because our brain can clearly tell us that the imitation is exactly that an imitation where Uncanny Valley starts to form is when we get too close to reality, but not quite close enough. If you were to plot realism against psychological acceptance, the curve moves steadily up and then falls off a cliff. That's the valley. And we'll get more into the science of this as the video goes along. So with the hard lessons learned from Polar Express, animation studios began test screening their films. When the final cut of Shrek was screened, the Princess Fiona character actually reduced children to tears. Children screaming in fear is not funny. It depends on the context. Fair enough. So to fix the Princess Fiona problem, the studio actually went back and made Shrek feel less lifelike in order to avoid traumatizing children. Computer games are also becoming more realistic and they can blur the line between virtual reality and actual reality. And game designers are very aware of Uncanny Valley and they work hard to avoid it. The term Uncanny Valley was first hypothesized in 1970 by Japanese roboticist Masahiro Mori, who identified that as robots became more and more human like, people found them to be more acceptable and appealing than their mechanical counterparts. But this only held true up to a point when they were close to but not quite human. People developed a sense of unease and discomfort. Take Shrek, for example. Most people remember this movie as a fun frolic through fantasy land. Hated it. Hated if you want. Most people loved it. Anyway, Shrek hit the sweet spot that Mori was hinting at in his hypothesis. When Princess Fiona looked almost human, she was scary as hell. But make her more like a cartoon. Suddenly she's likable. Did you say lickable? Likable. So if it wasn't for that test screening and the identification of characters that were landing in Uncanny Valley, Shrek might never have happened. We would have been fine without it. People love Shrek sheep. Moving on. Modern robotics provides the most context for Uncanny Valley, which makes sense considering the term originated from that field. We've all seen robots that are made to be as lifelike as possible, but we get uncomfortable when they move unnaturally or have kind of lifeless eyes. Very corpsey, corpsey, you know, corpsey, like a dead body. I know what a corpse is. I just like that thing in your closet you bought from the Internet. Oh, there's nothing in my closet. Oh, come on. You're on Amazon tracking the order like every 15 minutes until it got here. You kept staring at the window like a puppy. There's nothing in my closet. You know, if your neighbors saw that thing, they'd call the cops. There's nothing in my closet. Right, right, right, right. There's no room for anything else with all those skeletons in there. Engineers are very aware of Uncanny Valley, so they work hard to avoid it. Take Osimo, the Honda designed upright Android. Osimo is very human that it can walk, run, play soccer and do all kinds of other stuff. But Osimo doesn't creep us out. A primary reason that people love this robot and don't fear it is because while Osimo's designers have replicated a lot of human behavior, they haven't tried to replicate too many human physical features. Osimo is clearly a robot, a simple machine and nothing more. Now, it might seem counterintuitive, but something that acts human but looks artificial is much more agreeable than something that looks human and acts artificial. It is very nice to meet you. Much more agreeable, Asimo doesn't fall into Uncanny Valley, but put a human head on the thing. Suddenly it's a horror movie. Shrek is still worse. If we haven't earned your subscription today, I promise this moron is going to keep making these stupid videos until you hit the button. So is Uncanny Valley just a scientific theory or is it something innate in all of us? It appears as though it's something all humans share with varying degrees of severity, but it's been scientifically proven. Academics Maya Mather and David Reikling studied 80 real world robots and found a clear value effect in how much people liked and were willing to trust each individual robot. It was essentially a ranking exercise from least creepy to most participants were asked to rank which robots they would trust to invest one hundred dollars of their money. It turns out that the robots that were closest to human form, close to the valley threshold, they weren't getting a dollar. The participants didn't trust them, but the robots that just looked like robots, they were deemed more trustworthy. Real, not real. I don't know what that means. If scientists agree that Uncanny Valley does indeed exist, where's the valley? It's thought that Uncanny Valley might occur at the boundary where something moves from one category to another, in this case between non-human and human. Psychologists Christine Looser and Thalia Wheatley looked at mannequin faces and found the uneasy tipping point to where the mannequins looked too lifelike. Now, mannequins are definitely creepy, but we tolerate them in the normal context of using them to display clothes. But take a minute to think about a mannequin in a department store not wearing clothes. Are we going to address what's in your closet now? No, no, no, no, no. We've moved on. Many people find undressed mannequins disconcerting. Now imagine if they showed signs of life. Yikes. What Lucer and Wheatley showed is that there's definitely a point of no return where a robot or artificial human appeared to show actual human consciousness. We jump into Uncanny Valley when we believe a robot can reason, learn and feel like humans do. You know, I'm afraid of robots. Yeah. With artificial intelligence improving as quickly as it is, it's no wonder that this is a concern. A study by researchers Kurt Gray and Daniel Wegner found that robots were only unnerving when people thought they had the ability to sense and experience things. If people felt like robots couldn't reason on their own, they didn't seem as frightening. We fear artificial life is having human like consciousness without the control that humans have. Essentially, we worry that they're going to go all terminated matrix. I am giant. Right. I robot RoboCop Westworld Westworld again. OK. Metropolis. A.I. to step foot wide. OK, I got it. It becomes self-aware to 14 a.m. Eastern Time, August twenty ninth. Now, of course, Uncanny Valley only occurs if we can tell that these things are artificial. Once we cross the valley entirely and we perceive something or someone to be fully human, we then become more comfortable because we consciously or subconsciously perceive them as actually human, even if we know they're not. I prefer the term artificial person myself. Creating artificial life that makes the leap over Uncanny Valley is beyond our current state of engineering, but it's definitely on the horizon. Obviously, this brings up all sorts of anxieties into what's real or not. Then there's the old Blade Runner conundrum as to whether something is a robot or not. Is this testing whether I'm a replicant or a lesbian, Mr. Deckard? Just answer the questions, please. There's also a theory that robots and I could learn to cross Uncanny Valley on their own without human intervention. Now, this happens. Is that consciousness? Is that self-awareness? These questions start to veer into philosophy and existentialism. Please don't. We're not doing it now. Thank goodness. But we have a video coming up on consciousness. So hit subscribe and the bell and all the fancy buttons. You'll be notified when it's ready. I'd rather watch Shrek. Another trigger for Uncanny Valley is seeing behavior that contradicts expected emotion or lack of emotion altogether, which reads robotic. This is a unique theory. Angela Tynwell has led studies in virtual characters exploring these mismatches of emotion and behavior, and one study looked at the lack of surprise in a character's upper half of the face where the lower half had a reaction. And this caused all sorts of eeriness and led to these characters to develop similar mannerisms to that of psychopaths. Humans can recognize these facial patterns automatically. Take a look at someone who's smiling sincerely versus someone who's smiling insincerely. You'll notice that with a sincere smile, muscles of the entire face move, especially the eyes. When someone is being insincere, only the muscles of the lower half of the face move. There has been further research into the jarring experience of having animated characters reacting with mismatched features and emotion, all of which left some pretty horrible feelings in the minds of the subjects, even though they were obviously watching something animated. Differences between eyes and other features were especially bad. Turns out a happy smile and angry eyes is not a good look and pushes the unfortunate recipient of this expression headfirst into the valley. So what's the purpose of this reaction? Well, the answer is debatable, but it's generally thought of as a self-preservation response encoded in our DNA through evolution. Uncanny Valley helps us perceive threats by detecting insincere emotions and could be a way for us to avoid people we may perceive as physically or emotionally ill. It's pretty clear that we're getting closer and closer to inventing robots and artificial intelligence that are indistinguishable from human life. So how will we know we've built a bridge across Uncanny Valley? We'll know. Time to talk about that closet. Thanks for hanging out with us today. My name is AJ. That's hecklefish. This has been the Y files. Do you have any good examples of Uncanny Valley? Post them in the comments section so we can all have panic attacks. We're working hard for you. So do us a favor. Give us a thumbs up. And if we haven't earned your subscription today, I promise we'll keep making fun, brainy content until we do earn it until next time. Be safe, be kind and know that you are appreciated.