Understanding the Uncanny Valley Phenomenon

To better explain the uncanny valley effect, imagine this: it’s 2053, and you’re at the doctor’s office. After entering the examination room, a young woman in a white coat confidently walks in, greets you by name, and smiles warmly. She explains several possible causes of your shoulder pain, but when she touches you to adjust the joint, you notice her skin feels unnaturally cold.

That’s when you realize something is off. Her eyes have an unnatural glassy quality, and her gaze never aligns with yours. As she turns to prepare a machine, you hear the faint hum of mechanical motors, and you suddenly realize the truth: you’ve been examined by an android. She appears human, yet the subtle differences unsettle you. Welcome to the uncanny valley.

The concept of the uncanny valley was introduced in 1970 by Japanese roboticist Masahiro Mori, a pioneer in robotics. Mori suggested that as artificial humans become more lifelike, our affinity for them increases—until they near perfection, at which point small differences make us uncomfortable and our connection to them diminishes.

If these artificial beings were to become indistinguishable from actual humans, we would once again feel a sense of affinity towards them. In a graph depicting affinity against realism, the sharp dip in affinity creates a valley shape. Mori referred to this phenomenon as bukimi no tani, which loosely translates to 'the uncanny valley.'

Since the concept of the uncanny valley was introduced, it has evolved from a mere philosophical idea into an effect that has practical consequences. Computer-generated human characters are now a common feature in films, and engineers are continuously working on creating hyper-realistic androids, the term used by roboticists for human-like robots.

The day when humanoid robots assist you at the doctor’s office or even live in your home is getting closer every year, making how we interact with and react to artificial humans increasingly important. Let’s dive deeper into Mori’s theories and the factors that trigger the uncanny valley effect, exploring how creators can design artificial beings that avoid this eerie phenomenon—if such a thing is even possible.

The Origin of the Uncanny Valley

Masahiro Mori's initial explanation of the uncanny valley is relatively simple: We enjoy seeing artificial beings that resemble humans, but we experience an unsettling feeling when they come very close to being human, but with a few subtle imperfections. However, delving deeper into this concept reveals a lot more complexity, particularly when trying to fully understand what Mori meant in his original publication on the theory.

Mori's essay 'Bukimi No Tani' was first published in the Japanese journal *Energy* in 1970, but the concept of the uncanny valley didn’t gain widespread attention in the West until the article was translated into English in 2005. Even then, the essay wasn't fully translated until 2012 (the initial translation wasn't intended for publication), meaning our early understanding of the uncanny valley theory wasn't entirely accurate [source: Hsu]. For instance, the term bukimi could be more accurately translated as 'eerie' rather than 'uncanny,' although the 'uncanny' term is so widely recognized that it couldn't be discarded. Moreover, the term Mori used for our affection for artificial humans, *shinwakan*, is not easily translated. The original translation of 'familiarity' doesn’t quite capture Mori’s intent, and researchers have instead used the word 'affinity,' indicating that *shinwakan* blends both familiarity and likability [source: Kätsyri].

Mori placed various examples of human-like objects on his uncanny valley graph, including industrial robots, toy robots, and prosthetic hands. Many of these examples are specific to Japanese culture, such as bunraku puppets and Okina masks. While these examples are valid, they can pose challenges for Western researchers in terms of study and comprehension.

In conclusion, Mori proposed a possible explanation for the uncanny valley effect: He positioned corpses and zombies at the lowest point of the valley on the graph, suggesting that our discomfort with nearly-human simulations arises because they evoke the image of corpses, triggering our innate fear of death.

Mori's theory of the uncanny valley stemmed from his own ideas on how humans might interact with humanlike robots, rather than being based on statistical studies of human/robot interactions (which would have been difficult to carry out in 1970). However, his hypothesis sparked a wave of research into the phenomenon that followed.

Mapping the Uncanny Valley

To better understand the uncanny valley, let's explore some examples from both real life and pop culture. At the lower end of the graph, industrial robots are found—they don’t resemble humans and don’t evoke much affinity. An android like C-3PO from *Star Wars* would occupy the middle of the chart: His shape is humanlike, and he behaves like one, but his metallic body and robotic face make it clear he's not human. Still, we do feel some level of affection for him.

Further along the uncanny valley graph are computer-generated humans from Disney films like *Frozen* and *Moana*. While these characters clearly represent humans, their features are intentionally exaggerated to avoid appearing too lifelike. The success of these films shows that audiences connect with these characters, feeling a strong sense of affinity for them.

A notable example of an unsettling simulation is the computer-generated version of Tom Hanks in the 2004 animated film *The Polar Express*. The film's creators aimed to create a lifelike character, but the result fell short, leading many critics to describe the film as creepy or nightmarish instead of charming [source: Zacharek]. That eerie version of Tom Hanks? It's positioned near the bottom of the uncanny valley. According to Mori, the intensity of the uncanny valley effect becomes even stronger when these simulations are in motion rather than static.

When human characteristics—such as voice, proportions, and texture—are inconsistent in replicas, they cause discomfort. Mori’s hypothesis, which suggests that slightly flawed human replicas remind us of corpses and death, may have some truth, but it doesn't fully explain the complexity of the uncanny valley. The phenomenon is likely the result of several different reactions. Here are some reasons why nearly-perfect human simulations might unsettle us [sources: Hsu; Looser and Wheatley]:

Researchers have been diligently investigating the causes of the uncanny valley. Let’s explore some recent studies that have tested the uncanny valley effect and uncovered data about its potential origins.

Studies on the Uncanny Valley

A challenge in studying the uncanny valley effect is the difficulty of quantifying concepts like affinity and lifelikeness. Nevertheless, researchers have designed experiments to detect and analyze the uncanny valley effect, even attempting to offer mathematical explanations. One notable discovery is that the uncanny valley doesn't manifest in every study searching for it, and when it does, its intensity varies.

This implies that the uncanny valley effect is real but is triggered by specific factors, which is why it doesn’t appear in every study. For example, one study found that participants were more adept at distinguishing real from artificial humans when they focused only on the eyes (as opposed to just the nose or mouth), suggesting that perfecting the eyes is crucial for creating realistic human replicas [source: Looser & Wheatley].

Even something as simple as an unnatural pose or facial expression can trigger the uncanny valley effect, as demonstrated by research showing that people found humanlike virtual characters most unsettling when they failed to show proper facial reactions to being startled [source: Tinwell et al].

Replicas delve deeper into the uncanny valley when they attempt to 'trick' viewers into thinking they are human, rather than merely portraying highly realistic androids. A 2012 study revealed that people feel most uncomfortable when humanlike robots seem to possess minds and the ability to feel and perceive [source: Gray and Wegner]. Additionally, one study found that the uncanny valley effect only occurs when people observe faces that resemble their own ethnic group [source: Hsu].

The uncanny valley effect goes beyond just humans. One intriguing study observed monkeys' reactions to a variety of both real and artificial monkey faces. The researchers found that the monkeys exhibited a distinct valley when faced with the more realistic artificial faces [source: Steckenfinger and Ghazanfar]. This research collectively suggests that the uncanny valley does indeed exist, but it evokes diverse responses in different humans. This implies that overcoming the uncanny valley would be a challenging feat, as a humanoid robot that might pass for one viewer could still provoke discomfort in another.

Mori's approach to overcoming the valley? His advice was clear: don't attempt it. He recommended that roboticists keep their humanoid robots on the left side of the uncanny valley, emphasizing exaggerated features to enhance affinity and steer clear of the unease a more lifelike android might cause (similar to Disney's approach).

Other experts argue that navigating the valley is impossible because it’s not just a dip — it’s more like a wall. As technology advances and artificial humans grow more realistic, humans' ability to detect subtle differences also intensifies, making it harder to transcend the uncanny valley.

The uncanny valley isn’t confined to academic studies. In fact, there are real-world examples of figures that inhabit this eerie space. What familiar faces fall into the uncanny valley? Let's take a look.

Visiting the Uncanny Valley

The concept of creating objects that resemble humans dates back far beyond the idea of the uncanny valley. A prime example is Michelangelo's Pietà, an exceptionally realistic sculpture. For centuries, we've admired such works of art for their craftsmanship and the emotion they convey. But perhaps because these statues are static and often lack color, they don't try to mimic humans exactly and therefore don't elicit the discomfort typical of more lifelike attempts.

In contrast, consider the works of artist Ron Mueck: his sculptures of humans, painted with astonishing realism. Even though Mueck sometimes creates larger-than-life figures or characters with extraordinary features, his more ordinary scenes—such as two elderly women conversing or a couple lying in bed—can be unsettling. Mueck seems to intentionally provoke the uncanny valley effect in his creations. This intentional eeriness can also be found in certain Japanese horror films, such as the unsettling movements of humanlike figures in "Kairo" and "Ringu."

When Mori first introduced the idea of the uncanny valley, there were no real-world examples of highly realistic robots or computer-generated characters. However, technological advancements in robotics and computer graphics have made the uncanny valley an increasingly familiar experience. A significant example of this was the 2001 movie "Final Fantasy: The Spirits Within," which showcased ultrarealistic CGI characters. The aforementioned "The Polar Express" and 2007's "Beowulf" were also criticized for the unsettling quality of their nearly-human characters, further popularizing the uncanny valley concept in the U.S.

More recently, the Star Wars film "Rogue One" featured computer-generated versions of actors Peter Cushing (as Grand Moff Tarkin) and Carrie Fisher (as Princess Leia). These characters were generally received more positively than earlier CGI renditions, likely due to their relatively short screen time and advancements in animation technology that helped pull them closer to realism, reducing their presence within the uncanny valley.

Androids are no longer a distant dream from science-fiction movies, thanks to breakthroughs in software, materials, and electronics. Among the most lifelike androids are artistic creations like "Nova Phil," a highly accurate replica of science-fiction writer Philip K. Dick by Hanson Robotics. Others, such as the Actroid robots from Japanese company Kokoro, serve as technological showcases. These robots have been used as guides at public events and as telepresence devices, allowing people to interact remotely. A prime example is Hiroshi Ishiguro's android replica, which enables him to remotely teach university classes. The increasing realism of these robots edges them closer to the uncanny valley.

While it's unlikely we'll have android doctors anytime soon, this future is much closer than we might think, with decades rather than centuries ahead of us. As robots and androids continue to play a bigger role in our daily lives, the impact of the uncanny valley will become even more relevant.