Definitions Deconstructed
Navigating the Uncanny Valley
S. G. Lacey
One eye on the incredibly dark roadway further afield, the other tracks the incredibly bright cellphone display attached to a holder on the center console. Where the hell is this hotel? Or at least any sign of the town this establishment is supposedly located in?
Suddenly, a moving blur appears in my fatigued peripheral vision, traveling towards me, on the edge of the roadside. Seconds later, before I can even engage my faculties to slow the vehicle, this apparition darts across my path, a non-descript grey shape briefly silhouetted in the headlights.
Too dark to be a spotted cow, potentially escaping from a nearby field. Too tall to be a mule deer, undoubtably common in this wooded area. Too fast to be a moose, who typically lumber along indifferently.
Plus, this creature appears to positioned vertically, potentially on a pair of legs, as opposed to horizontally, on all fours, as is typical for mammals of the forest. This mode of movement is a decidedly human trait. A bear maybe. I’ve never seen one, so have no frame of reference. Maybe my tired senses are simply betraying me.
I’m making this aggressive road trip spur of the moment. Which explains the complete lack of planning, and multitude of issues encountered along the way.
I’m trying meet up with my new boyfriend, for the first time in person. If I ever get there.
I had a sociology class test earlier this Friday morning, which should be illegal in a collegiate setting, so wasn’t able to get on the road until 3 PM. The drive, which supposedly only takes 7 hours, based on my online mapping research, is now on its 10th hour, due to a combination of heavy thunderstorms, road construction, and navigational gaffs.
At least the recent odd animal, I hope, experience on the roadway a few miles back, has kicked my previously sluggish mind into a high gear of alertness. Granted, this still-developing cranial matter perpetually wanders.
The odd, upright blob I spotted gets me thinking about sports mascots. Understandable, since the main purpose of this random journey is a football game. And some frisky fun, hopefully.
I’m a junior at Michigan State. Our squad is on the road this weekend against our developing nemesis, Penn State, in the constantly evolving Big 10 conference. Which happens to be my current fling’s graduate school posting.
We definitely have a better mascot, a majestic Spartan warrior. An incredibly muscular man, provided with all manner of supplies: sturdy sandal boots, ab-enhancing torso armor, metallic wrist guards, and a fancy helmet adorned with feathery plumage. Granted, the full emerald green garb is a little goofy, and this soldier’s name could be more original.
Still, I’m headed to a higher educational establishment with a goofy lion caricature, which utilizes poop-brown, coarsely-matted, synthetic fabric, combined with awkwardly-added ear, mouth, and tail elements. Clad in nothing but a navy and white stripped scarf, which looks like it should be part of prison garb.
Half boy, half beast, zero intimidation. The folks here in rural Pennsylvania are so uninventive, or conceited, that they named the local township State College.
At this sluggish rate of travel, I’ll be lucky to arrive there before the tailgating starts. I was originally planning to surprise my boy tonight, but considering my late ETA, and lack of prior communication, it would be risky and challenging to try tracking him down on a Friday night. Frat boys will be frat boys.
Hence the hotel accommodation, which I’ve been perpetually scheming on, and audibling on, throughout this entire slow drive.
Time to call my target lodging, booked over the internet on my cell from a rest stop a few hours back, and let them know I’m going to need a late check-in.
I hate these automated phone services. They never recognize my voice.
“Reservations!”, I repeat for the 3rd time, in a very slow and elongated cadence. Suddenly, the elevator mood music which relayed the fact that I was on hold the past 5 minutes disappears, replaced by an electronic utterance.
Amazingly, even though the voice on the other end of the line is clearly computer generated, it still has that generic Indian accent typical of outsourced overseas call centers.
It’s almost worse to know that Western society has stereotypically engrained such customer service roles to a specific region of the globe. Plus, this technological advancement is no doubt developed in the same distant region, by a more educated cohort, to cut costs and streamline operations.
Global geopolitics aside, this is not a very functional virtual assistant. The programmers have hidden the artificial part fairly well, but the intelligence characteristics are still highly lacking. At least in a pair of key areas for distressed travelers: voice recognition and empathic support.
I was making more progress while on hold. Maybe the algorithms aren’t calibrated for rapid yelling from the subject, with the dull roar of a breeze from my fully lowered driver’s side window. I need to stay awake somehow.
Fortunately, before I get frustrated enough to toss my cell out the opening into the dense brush lining the road, this wall of foliage breaks, offering up a fenced-in field. More importantly, twinkling lights in the distance suggest some form of development is nearby. My nighttime sanctuary is near.
I offer up a terse and completely disingenuous “Thanks.” to my fake Hindi helper, who seems indifferent if I hang up, or if they provided any help whatsoever to my current challenge. Still, even as I press the red circle on my screen to end the call, I feel a dinge of remorse for my vulgar behavior.
Which I quickly dismiss. There wasn’t a real human on the other end of the line anyways.
Rolling through town at a crawl, I finally spot my destination. The Happy Valley Hotel. What a journey it’s been.
At least the place is easy to identify, considering the gaudy branding which identifies this establishment. The lighted signage, shaped in the profile of twin wooded mountain peaks, with a mock blue river flowing between the cleavage, is very reminiscent of the country terrain I assume I just drove through in the dark over the past few hours.
The words “HAPPY” and “VALLEY” are displayed at an angle in bulbous capital letters, using a white font that’s easy to read against the emerald green background.
But the most visible, and awkward, element, of the neon impingement sits below this naturescape. A pair of caricature faces, whom I can only assume are the proud owners of this facility. Despite the contracted artists best efforts, this couple’s appearance is hideous, and oddly similar.
Stringy dark hair, one balding up top, the other a frizzy mess. Rounded faces with plump cheeks; neither has missed a meal in a while. Toothy grins, generally filled in, but with a combined missing tally approaching a half dozen. This duo must be big hockey participants.
But the ears are what catch my attention, as these elements are huge, completely out of proportion relative to the rest of the anatomical features. Is this an accidental gaff by the sign’s designer, or is there a more subtle message that this pair is listening to everything which happens in their business.
Apparently, I’m about to find out.
Screeching into the gravel driveway, on what feels like 2 wheels based on my excessive rate of speed, I make a bee-line for the space closest to the entrance. Which is empty, along with the entire parking lot. Must be a slow night.
Talking to a real human will be much easier than the interaction which just went down during that awkward phone chat. Getting out of my car, I slam the door in fatigued frustration, and stride purposefully towards the low-slung building.
Entering through the revolving door carousel, an absurd and clearly unnecessary splurge at this budget business, I enter the small lobby, and immediately turn my gaze towards the front desk, behind which service personnel would traditionally be located. No friendly counterperson’s smile matches my icy gaze. What kind of a bootleg operation is this?
With no live individual materializing at this ungodly hour, it appears I’ll have to utilize the electronic kiosk positioned in the corner. Great, more marginal technology. Another detrimental element of my late arrival.
I typically try to avoid these automated systems, especially at grocery stores, where each produce item must be individually documented and weighted on the sensitive scale, or car rental lots, where the required ID documentation inevitably doesn’t register correctly on their grimy scanners.
How hard can it be to check-in? I guess I can brave the technological tower.
3 minutes later, I’m already regretting my decision. Rather than being a simple touch screen platform, I’m forced to engage with a person, albeit in digital hologram form.
The movement of the plump mouth is completely out of sync with the verbal prompts provided, and the obviously disproportioned hand gestures, which are meant to guide my inputs to the pedestal, but are so generic as to be completely useless.
I’m soon ready to punch the pad, as opposed to gently touch the tablet.
Apparently, my anger management issues are heightened when I’m tired and hungry. Not a moment too soon, my disjointed sequence of button pressing and spoken utterances has been interpreted. A keycard drops out the bottom on the machine with a clink, that falls somewhere between plastic and metallic on the auditory scale.
My aggravating avatar floating in the air gives me a wink, gestures downward, apparently towards the distribution slot, then utters a slurred “Enjoy your stay.” before the rendering disappears. Good riddance.
Hopefully this experience isn’t what the future of customer service looks like. But I’m too fatigued to submit a complaint right now. Maybe tomorrow, I’ll log onto the corporate website and give them a piece of my mind.
I check the numerals recently etched into my access card. “143”. At least I’m on the first floor, so it shouldn’t be a long walk.
As I meander down the hallway to my assigned room, distracted by analyzing the numbers on each successive door, I nearly step on an encumbrance moving slowly along the ground. Startled, I look down, expecting to find a discarded towel, or even worse, a cockroach, given the disheveled condition of this facility.
Instead, I’m met with a shiny metal disk, about the size of a dinner plate, gliding across the worn carpet. The top of this circle is adorned with a pair of exaggerated black eyelashes over bright blue eyes, and a broad smile, bright white teeth inside pink lips.
Clearly stickers applied by some comedian on the janitorial staff, but the achieved visual effect, the combination of shapes, and their relevant placement, immediately draws reference to a face, even in my sluggish mind.
My surprise quickly turns to intrigue; I stop in the middle of the lane, and watch mesmerized as this mobile creature meanders back and forth, occasionally bumping into walls, then reversing course, and proceeding onward at a plodding pace.
This path seems completely arbitrary, but I sense that given enough time, this cleaning robot will cover every inch of exposed floor. Unless an aggrieved patron like myself throws the machine into the trash bin first. Fortunately, the gaping grin beaming up from below is quite disarming.
Eventually, I move on, and finally find the target labelled door. Figures, it’s at the farthest end of the corridor. After all these robotic interactions, I’m happy to reach the safe sanctity of my room.
Considering my bad string of luck over the past few hours, I half expect the digital lock to be broken. However, it yields quickly, as displayed by the green dot on the exterior panel, as soon as my electronic key card reaches the required proximity. I’m finally in.
Flashes of light emanate from the dark interior space as soon as I open the door. Already on edge, I enter slowly, but the source of the stimulation quickly becomes evident. And is completely innocuous.
The flat screen television is on, displaying my name as part of a clearly generic welcome message, interspersed with still shots of various hotel amenities. Which clearly must be curated from stock footage, based on what I’ve seen of this dump thus far.
I spend a few minutes perusing through the channels with the remote. An addition which afflicts many of my generation, I can’t help but explore the available offerings on any new electronic device.
As with most cable TV systems, there’s no shortage of options, none of which are good. I rapidly navigate past all manner of content, pausing just long enough to determine each show’s not worth wasting my time.
A children’s cartoon, characters so lifelike that it’s a little bit eerie to watch. Yet another reason why I haven’t birthed any offspring to date.
Some sort of medical drama, surgery occurring on inert actors, while detailed simulated overlays provide internal insight on the invasive procedures. No thanks, I don’t even like eating meat with bones present.
Japanese anime, clearly adult themed, which aligns well with my preference for fictional fantasy literature. However, the purple and pink hues, on skirts and stockings, are a little to girlie for my own emo goth leanings.
An action movie, utilizing so much CGI that it’s impossible to differentiate real from rendered, scripted from simulated. This could be a viable option, but if I get roped in now, I may sprawl out on the bed and never wake up.
Enough distractions, time to get settled in. Moving to the bathroom, I drop my cosmetic bag next to the sink, then turn to take relief after the lengthy journey, and nearly wet myself. Leering in the corner is a creepy creature.
A full size body. Initially aghast, I’m slowly defused as I realize the nature of this stalker. Fortunately, this item is an inert mannequin, as opposed to a potential rapist.
After my recovering from my original fright, I move closer to examine this odd inclusion to the restroom.
The body color is pale, not dissimilar to my own flesh tone. Without considering if I should, I reach out and touch the forearm. The surface is rigid and cold, a far cry my soft and warm skin. Likely some sort of plastic construction.
Instinctively, I glance at the key elements of this nude figure, that should help confirm sex. Which doesn’t provide much clarification. The groin is smooth, with a slight, featureless bulge. The chest is similar, marginal lumps non-confirming of muscular male or flat female.
Not able to turn away, curiosity overwhelming confusion, I continue my examination. This caricature is missing a few key elements of the human anatomy. No feet. No hands. And most importantly. No head.
Still, even without eyes, or a face, it feels like this statue is watching me.
I have no idea what this dummy is for, or why it’s here. Maybe offered up to patrons as a means of dewrinkling clothes, or examine potential wardrobe combinations. I’m not heading to a wedding, or funeral, tomorrow, so plan to keep my basic outfits safely stowed in my luggage.
I still need to pee, even more so now, but am having a hard time relaxing in front of this androgenous creep. Hefting the weighty object around until what I perceive as the front of the figure is facing directly into the wall, I’m finally able to do my business in peace.
This place is just weird. I’ll be happy to chug the can of wine in my backpack, pass out, and hit the road bright and early in the morning.
Provided no more disturbing distractions, lifelike or otherwise, invade my evening. The Happy Valley Hotel has some real issues in my book. I can only image what kind of dreams will materialize, in a foreign bed, sleep deprived after a long day on the road, with many awkward semi-human engagements recently cemented in my mind.
The string of random occurrences experienced on this journey are uncanny.
Definition:
Uncanny valley = Theorized relation between the human likeness of an object and a viewer’s affinity toward it. [REF]
Deconstruction:
The hypothesis of the “uncanny valley” originated with Masahiro Mori, a Japanese automation engineer, in 1970.
For his original published thesis, the phenomenon was referred to as “bukimi no tani genshō” in the native tongue. The fundamental premise was based on enabling robotic technology adoption in society, which was Mori’s university specialization and lifelong career pursuit.
The entirety of Mori-san’s extensive research paper can be summed up in the single labelled graph shown below. [REF]
Most of the translated objects listed on this formative plot are commonly known. Aside from one decidedly Japanese item.
“Bunraku” is a traditional form of Japanese theatre, originating in Osaka during the 17th century, which includes puppeteers, vocalists, and musicians. These delicate dolls are carefully crafted, with elaborate traditional garb, and delicately painted faces.
Aside from their small stature and slightly jerky movements, depending on the viewing perspective, a well-executed Bunraku puppet performance can be nearly indistinguishable from natural human movement. Hence, it’s chosen placement on the literal right side of the uncanny valley.
One of the most interesting findings from this seminal treatise was highlighting the difference in human perception between static and moving entities. Like encountering an inert cadaver on a shiny morgue table, as compared to a crazed ghoul traversing the barren landscape, both in the virtual video game world of course.
The fundamental theory is that dynamic motion is more lifelike, and thus more worrisome for viewers, as evidenced by the deeper trough for the dotted line in the diagram. Considering the proliferation of zombies on modern entertainment platforms, society is craving creepy.
The specific uncanny valley terminology didn’t come into public consciousness until it was dubbed by British art critic Jasia Reichardt in 1978. This verbiage has stuck through to the current day despite not being completely accurate from a literary standpoint. There are many synonyms of “uncanny” which are more common words in the English lexicon.
This relatively obscure research by Masahiro Mori came back into vogue when the entire paper was fully translated into English in 2005, with the help of technology, for both interpretation and dissemination.
The revulsion postulated by Mori is severe and steep, with increased positive feelings from individuals for an object which is either less, or more, lifelike. Hence, the valley terminology.
When caught in the chasm, observers’ responses can manifest in many ways: disturbed repulsion, detracted realism, decided reproach. Beyond the individual dismissal issues, from a broader societal standpoint, such distrust can hinder the future development of robotics technology, in both the physical and virtual realms.
The key takeaway from Mori’s paper is the valley element, a specific zone on human likeness that is ironically disliked. However, his research acknowledges that as simulation and automation capabilities gets increasingly acute, people will again start to the enjoy these conceptions, with the logical terminus where such androids are essentially indistinguishable from their creators.
While no actual interview polling was done for Masahiro Mori’s original report, in subsequent years, various researchers have tried to tackle this topic in a more scientific survey manner.
A 2009 study presented monkeys with a trio of picture options: a cartoonish caricature, a realistic rendering, and a physical photo, all of their mammalian species. The primates spent the least time, as tracked by eye movement, perusing the digitized doppelganger work of their relatives. Score one for the valley boys, or baboons, in this case.
Similar studies with actual human subjects in recent years have been slightly less convincing.
A survey with proficient video gamers, examining how verbal utterances and facial expressions of virtual characters was interpreted and reacted to by participants during live gameplay. A collection of robot facial images presented to common citizens, ranging from explicitly mechanical metal, to near-perfect skin craftmanship, while their own facial responses were monitored. Children given a range of toys, with differing levels of realism, in terms of both visual and tactile accuracy, then their playtime choices were monitored.
In all these instances, the uncanny valley concept was potentially observed, but not explicitly confirmed. Like any psychological premise, validation itself is subjective. Especially when digital renderings and daily reality merge.
The uncanny valley effect can occur in all manner of modern entertainment scenarios: advanced robotic creations, video game simulations, CGI movie characters, and even physical mannequins. The realism can be enhanced, or minimized, depending on the context within which each simulation is encountered.
Mori’s research doesn’t comment on natural individual preferences with regards to human attraction or aesthetic preferences. This is a completely different field of research, which may be relevant in the future, as certain physical traits are perceived as pleasurable or disarming by various cultures.
Everyone has experienced the scenario where there’s something odd about a given picture, or situation, not just those involving fellow humans. As a result, the uncanny valley phenomenon may simply be an extension of intuition, the innate trait of pattern matching which has keep people safe since the era of cave folk.
Not surprisingly, many of the most realistic android robots, and thus the most eerie to watch, have been developed in Japan. This culture has always had a penchant for technological advancement and simulated experiences, which makes them perfect practitioners for lifelike automats.
With names like Telenoid R1, Sanya, Alter 3, Geminoid Hi, CB2, and Actroid-F, the descriptions don’t exactly facilitate cordial conversation. Several of these offering spawned from the University of Osaka, a robotics innovation hub where many experts in this unique space learned their skills. Sounds like a pretty creepy campus to wander around on.
Some of the most intriguing examples demonstrating the uncanny valley phenomenon to the masses have come from the film industry over the past few decades, as computer processing technology has advanced, an electronically-rendered characters came into vogue.
In many cases, the challenge for programmers is not the visual realism of their creations, but rather the intangible elements which make humans, well, human. Empathy. Anger. Intuition. Remorse. Bemusement. Fear. Love. These feelings, with the resulting natural facial features and bodily gestures, can’t be easily converted to computer code.
Here are some interesting examples of entertainment companies overstepping the boundaries of societal norms, or adjusting content based on preliminary screening reviews, due to the uncanny valley effect.
Tin Toy (1988): This was one of the first short films released by Pixar, using their innovative and proprietary PhotoRealistic RenderMan software. The lead infant character, Billy, was fully digitized, performing standard baby activities in an innocent, but disconcertingly mechanized, manner. Not to mention the slightly disproportioned facial feature layout, and oddly shaped appendages, a physical element of the anatomy which continues to plague literal digital artists to this day. This production was a major unlock in terms of what could be simulated with computers, and thus spurred the film industry to consider potential uncanny valley issues more deeply moving forward. [REF]
Adventures of Tintin (2011): This full-length movie, based on the famous Belgian artist Hergé’s comic strip series, had the potentially to be a blockbuster. The realistically rendered action adventure was made using motion capture technology with live actors, then a generous dose of digitization on the back end. Unfortunately, the product tried to merge original stylized artistic elements with features which were too photo realistic. This is one of many films where a large production budget, allowing for substantial computer-generated imagery resources, combined with an ambitious director, ended up putting too many cooks in the kitchen. Despite being an incredibly popular comic franchise worldwide, the inaugural movie was a flop, no sequels having been produced to date, despite the clear multitude of additional existing content. [REF]
Sophia (2017): Though not explicitly in the film realm, this entertainment phenomenon made her TV debut challenging host Jimmy Fallon to an awkward game of rock-paper-scissors on the Tonight Show stage, then graced the cover of Cosmopolitan magazine in full bedazzled garb a year later. This AI-powered humanoid created by Hanson Robotics, a leader in the space, may be representative of a not-too-distance automat future. Sophia’s claim to fame is striving to be a fully sentient being; for now, this logical lass is simply the embodiment of cutting edge robotic artificial intelligence. For some luddites, this future state of anthropoids and automats cohabitating is obscene, while for uber-optimists, such mingling represents the inevitable future state of society. Only time, and technology, will tell. [REF]
Cats (2019): This was a digitized big screen adaptation of famous Broadway production. The animal-human hybrid faces were polarizing for both old and young viewers, eliciting a range of reactions from amusing to awkward to awful. There was also a disconnect in the movement patterns of these simulated creatures, blending feline and anthropoid traits. This viewership response demonstrates that the uncanny valley concept isn’t explicitly limited to the human form. This film flop also highlights the additional realism challenges incurred when portraying dynamic elements, as Mori noted in his original paper. For those folks who already don’t like housecats, this offering was definitely a non-starter. [REF]
M3GAN (2022): This recent horror film release, which involves a creepy AI robotic toy, would have been Mori-san’s research dream, or nightmare. Advances in robotic technology, motion-captured movements, and post-production CGI, help this film potentially cross the chasm to full realism at times, aside from the disturbing and satirical plot. Utilizing a child actor stunt double, combined with rendered overlays, and a heavy focus on emotional accuracy, this movie returned the uncanny valley concept to the mainstream big screen. There are several meta elements to this offering, including the lead character being a robotics engineer, and undertones about the effect of technology on children, for a release which was very popular amongst tweenagers at the theatre. While decidedly dystopian, this movie does raise some valuable questions about how humans and their lifelike robotic counterparts will interact, both physically and emotionally. [REF]
Masahiro Mori’s paper was released over half a century ago. Interestingly, there’s still no concrete scientific evidence or experiential polling to support any of Mori’s fundamental claims. Yet, this terminology is very relevant in a broad range of modern-day fields, from robotic science to artistic film.
Technology, especially with regards to simulating the human form in both the physical and digital realms, has come a long way since then. Computers were just in their infancy in 1970. Yet the uncanny valley concept is arguably even more important now than the decade it was conceived.
It will be interesting to see if society’s perception of, and aversion to, humanoid creations changes with further advances in technology. Robots used to be liked until they got too real, then these characters turned awkward. Now, consumers of content are frustrated if fake figures are less compelling and competent than their human counterparts.
Is it possible that processing and rendering capabilities have gotten so advanced that they’ve surpassed our own comprehension? Or are visual effects still just not quite up to snuff relative to numerous millennia of homo sapiens experiential adaptation and sensory refinement? Science would suggest the latter.
The real question is, are scientists and engineers designing around this perceived visual valley, dumbing down graphics which could potentially be much more realistic?
Currently, there’s a delicate balance for designers; presentation either needs to perfect, or purposefully processed. The explosive trajectory of AI, specifically video creation through replication, will speed up this process. These are truly interesting times to be employed in the digital real.
Is the uncanny valley a chasm or a creek, and how will technology advancements influence this assessment?
To be determined, but it’s already evident that civilization will forge across this gap, with tenacity, curiosity, and ambition in an uncanny risk-seeking way that’s uniquely human. Such greedy gumption doesn’t always go great.
Details: