Uncanny Valley Explained: 3 Uncanny Valley Examples

The uncanny valley is a phenomenon in which a human being experiences a negative emotional response to a non-human figure, object, or entity that possesses a human likeness. The underlying idea is that humans are comfortable with other human-like objects or entities up until a certain point. Once the object or entity looks too human despite its artificialness, it suddenly becomes “uncanny” and takes on a disturbing level of eeriness to the viewer. The character would elicit less of a negative response from humans if animators or developers were to dim or exaggerate its anthropomorphic features.

The uncanny valley phenomenon is most present in fields like robotics, animation, or video games, which rely on animators and developers creating human-like characters. Here are some examples of this peculiar valley of eeriness.

1. 1. Robotics: Robots that too closely simulate human behavior and experiences can be eerie to some people. One of the biggest examples of this is artificial intelligence (AI), which sometimes mimics human life as closely as possible. While a robot with a robotic voice meets peoples’ expectations, one with a human voice blurs the lines and may give someone the sensation that the character is untrustworthy.
2. 2. CGI animation: CGI animation in video games and films are prime examples of the uncanny valley effect. Computer technology is sophisticated and can often realistically portray human behavior and features. However, there are limits to how believable these characters appear: Animated films like Final Fantasy: Spirits Within (2001) and The Polar Express (2004) showcased strong human realism in their characters, which had the opposite effect. Some audience members felt unable to enjoy the movies due to the innate eeriness of the artificial human characters.
3. 3. Digital assistants: Digital assistants typically rely on audio alone, which is less likely to create the uncanny valley effect. However, certain companies give their digital assistants a human face, making them more likely to become subject to the uncanny valley effect. Many of these digital humans have a slightly delayed response when it comes to showing emotion on their face—like raising their eyebrows, moving their foreheads, or moving their mouths—which can arouse suspicion in the viewer.

There are a couple of main theories as to why the uncanny valley effect happens, and why it elicits a negative response from people. One theory notes that people fear the idea of automation and “robots taking over,” or not being able to tell the difference between human beings and AI. The theory asserts that the closer robots come to successfully mimicking human form and behaviors, the likelier they are to replace them.

Another theory is that the uncanny valley creates uncomfortable inconsistencies that make the categorization of an object difficult. Humans judge these artificial characters by human standards rather than robotic ones. Realistic robots and computer animations may have slightly different facial expressions than humans do. For instance, their facial movements may not match their words, or they might be slightly delayed. The viewer experiences a mismatch in expectations, which indicates that the character isn’t human enough for them to trust or like.

Here is a brief overview of research related to the uncanny valley phenomenon, beginning with the roboticist who gave the effect its name.

• Discovery: Japanese roboticist Masahiro Mori first coined the term “uncanny valley” (“bukimi no tani genshō”) in the 1970s to describe the way people feel about human-like robots. This robotics professor observed that people liked his robots the more human they looked, but only up to a particular point. This “valley” is where the person’s affinity for the robot suddenly dips, eliciting feelings of fear, disturbance, or anxiety. However, Charles Darwin had earlier noted an occurrence of the uncanny valley in 1839, in his book The Voyage of the Beagle, wherein he describes the face of a trigonocephalus viper and its resemblance to human facial features, resulting in a degree of disgust.
• Artificial intelligence research: Studies at the University of Cambridge revealed that participants who viewed different types of AI, such as android, mechanical, and humanoid robots, experienced increased blood flow in areas of the brain associated with evaluating the likability of visual stimuli. This brain activity was consistent with the uncanny valley hypothesis. Participants were more comfortable with robots that had a more human-like appearance, rather than those that were too close to the human/non-human boundary. This activity occurred primarily in the visual cortex and the ventromedial prefrontal cortex, which are both responsible for deciphering visual images and perceptual differences, as well as detecting human interaction—important components of interpreting social cues.
• Testing Mori’s theories: There is other evidence that purports that the uncanny valley effect may not occur at a specific spectrum of human faces and behavior, but will vary depending on a person’s background, culture, or age. American roboticist David Hanson Jr. disagreed with Mori’s theories that entities appearing more human-like will always elicit disgust or a negative reaction, and that better design or implementation of cartoonish features can help flatten the valley. Karl MacDorman, an associate professor of human-computer interaction at Indiana University, also viewed it as less of a spectrum and more of a way to gather observations about human realism. Associate professor Christoph Bartneck from the University of Canterbury in New Zealand believes that it is more of an “uncanny cliff,” and once the viewer falls off, there is nothing that can get them back.

The uncanny valley effect may have heavy implications for the future of robotics and android science. If technology cannot advance AI to the point of consistent comfort in lifelike robotic behavior, people may reject these advancements and stall meaningful technological progress due to simply being unnerved by robots attempting to mimic human form.

The phenomenon may also affect the future of prosthetic limbs. A human may find the appearance of a prosthetic hand or foot convincing at first, but it could later elicit feelings of unease. This negative reaction to life-changing technology could impact those with disabilities if they fail to create productive human/robot interaction, adversely affecting the funding and implementation of these tools.

Animators, developers, and roboticists can avoid the uncanny valley phenomenon by shifting the entity to the other side of the valley, or further away from it entirely. Giving the character, figure, or object either a more human-like, cartoonish, or robotic appearance can decrease feelings of discomfort or distrust in the viewer. For instance, some animators create characters with exaggerated facial proportions, such as lower mouths or large eyes, to deliberately avoid the uncanny valley effect.

Avoid mixing human and non-human elements in virtual characters or lifelike dolls to help create less of a discrepancy between visual cues and expectations. For instance, photorealistic textures should only occur with humanistic facial proportions rather than exaggerated facial features.

Get the MasterClass Annual Membership for exclusive access to video lessons taught by the world’s best, including Will Wright, Paul Krugman, Stephen Curry, Annie Leibovitz, and more.