What Is The Uncanny Valley Effect? Psychological Explanations And Examples

When reflecting on our relationships with technology and artificial intelligence, the uncanny valley effect may be a relevant concept. 

This hypothesis proposes that increasing human likeness in robots and other artificial characters does not necessarily appeal to humans. Alternatively, humanoid robots and other objects with human realism may lead to eeriness, unease, or other negative emotional responses.

Researchers continue to gather data to support the uncanny valley hypothesis, a fascinating research area in our technology-driven world. In this article following information, we will explore psychologists’ definitions of the uncanny valley and possible explanations for this phenomenon, along with five examples of this effect and strategies to avoid the uncanny valley. 

Defining the uncanny valley phenomenon

The uncanny valley was originally defined by Japanese roboticist Masahiro Mori, who published an article in 1970 about human beings’ responses to human-like robots. In his essay, Mori describes the uncanny valley as a “descent into eeriness,” evidenced by a graph of how humans perceive robots.

Through this graph and subsequent hypothesis, Mori makes a connection between the human likeness dimension– defined in terms of physical similarity to humans–and humans’ emotional experience of human-like robots and computer-generated characters, or avatars.

The “descent” into the uncanny valley is characterized by our responses to robots, which may shift from empathy to revulsion as we struggle to make sense of robots with human characteristics that are almost lifelike–but not quite. According to Mori’s hypothesis, feelings of affinity or likeability may transform into discomfort, eeriness, or even repulsion in response to artificial intelligence that seems “too” human.

Modern descriptions of the uncanny valley effect usually stem from the original concept proposed by Mori, though technology has developed considerably since the publication of his essay. As a result, depending on your access to and usage of technology, you may encounter examples of the uncanny valley semi-regularly.

What causes the uncanny valley?

In robotics, psychology, biology, and other fields, scientists continue to study the causes of the uncanny valley, which are complex.

Human neurology

The uncanny valley effect may be rooted in human neurology, as evidenced by a 2019 study of the neural mechanisms behind the uncanny valley. In this study, participants performed two tasks, both of which involved viewing and responding to questions about robots with varying levels of human likeness.

During the tasks, the neuroscientists measured higher activity in brain regions associated with social cognition, most notably two parts of the medial prefrontal cortex. It is in these regions, where the researchers found evidence of positive reactions to real humans and negative reactions to artificial figures based on functional magnetic resonance imaging (fMRI). 

Cognitive mechanisms

Cognitive and neurological mechanisms are closely related, but cognitive psychology focuses on the human mind and behavior. At the same time, neurologists study brain structure and the activation of certain brain regions during specific tasks.

In discussions of the uncanny valley, some cognitive scientists point toward the categorization difficulty hypothesis as a possible explanation for humans’ emotional responses and behaviors in the presence of lifelike robots. When people are asked to categorize an ambiguous stimulus as human or non-human, this hypothesis proposes that this task can lead to a negative emotional state.

Human psychology

Psychologists complement these neurological findings with their own hypotheses and studies of the uncanny valley. One psychological explanation comes from the perceptual mismatch theory, which suggests that inconsistent features–like robotic eyes or a mechanical mouth in an otherwise human face–trigger feelings of eeriness and creepiness associated with the uncanny valley. 

Evolutionary causes

Humans may express a natural, evolutionary aversion to human-like robots. This theory is supported by studies of adult macaque monkeys, who expressed preferences for real monkey faces and unrealistic monkey avatars over realistic monkey avatars. 

Because monkeys showed a similar aversion to realistic avatars of their own species, some scientists hypothesize that humans evolved some type of module for disgust or attractiveness in response to almost-lifelike avatars. 

While a growing body of studies backs these possible explanations, more research is needed to explain our behavioral and emotional responses to artificial beings with human qualities. 

Uncanny valley examples

As we venture further into the 21st century, lifelike robots and other artificial characters have become increasingly common–and convincing. The following examples highlight the effects of the uncanny valley and the potential for this experience to become more widespread as technology progresses.

Sophia

In the growing class of robots with human forms, Sophia is one of the most famous examples. Sophia was designed and released by Hanson Robotics in 2016 and is equipped with advanced-inducing abilities, including natural language processing and speech. Sophia can also perform complex emotional expressions, visual tracking, facial recognition, and other forms of human behavior. 

In an ongoing ranking of the “creepiest robots” by the Institute of Electrical and Electronics Engineers (IEEE), Sophia actually ranks lower on the list of comparatively creepy androids. Nonetheless, Sophia has received ample news coverage as an example of artificial intelligence with a human form and advanced emotional abilities. 

Telenoid

Telenoid is a robotic communications device. A person can connect to Telenoid over the internet, which allows the robot to transmit the human voice and mimic facial expressions and head motions.

According to the developers of this device, Telenoid is comparable to “a mobile phone shaped like an alien” with a bald head, doll-like face, and stumps in place of human limbs. While this simple design, described as a “minimalistic human,” is meant to facilitate human interactive communication, some people are hesitant to engage with Telenoid due to the creepiness of these features. 

Cats

The uncanny valley doesn’t just apply to robots. After watching the 2019 version of the musical Cats, many viewers reported feelings of unease in response to the feline characters. 

In this live-action reboot of the original musical, the cat-inspired characters adopted a more human appearance. In some reviews of the film, critics argued that the live-action characters were unconvincingly human, and the awkwardness of their facial expressions made it difficult for viewers to empathize with the characters and the overarching storyline. 

Geminoid HI-1

Geminoid HI-1 replicates a real human being, Hiroshi Ishiguro, who is also the robot's creator. Like Telenoid, this “android copy” of Ishiguro can reproduce the voice and head movements of the operator. 

This robot was first unveiled in 2006, and Ishiguro and his team have since used the robot to study human-robot interactions and “what it means to be human,” according to the IEEE. 

Diego-san

Diego-san is modeled after a 1-year-old baby. At first glance, the oversized features–including a large head and all the hardware to support Diego-san’s complex facial expressions–seem to conflict with the lifelike qualities of the face itself. 

Despite the potential creepiness of Diego-san, developers recognize the value of this robot to study cognitive and social development in infants. The latest version of Diego-san, activated in 2013, features high-definition cameras in the eyes and 27 “servo motors” that control its highly expressive face. 

Can we avoid the uncanny valley?

As roboticists and other scientists develop new, increasingly human versions of various robots, you may wonder: are there any significantly negative consequences of the uncanny valley, and if so, can we avoid these outcomes?

To date, the answers to these questions remain open to individual interpretation. Some argue that the uncanny valley is actually a positively good thing and that the associated feelings of discomfort remind us to stay curious about unfamiliar technologies and think critically about our consumption of media, which could include videos, pictures, and other content that may be influenced by artificial intelligence.

At the same time, some scientists are developing robots and avatars with the purpose of minimizing or avoiding the experience of the uncanny valley. Some of their efforts include: 

• Avoiding “inconsistent realism”: the combination of atypical features (such as skin textures or eyes that are not typically human-like) with high levels of realism. 
• Using subtle features in the eyes to express a character’s intention, behavior, and well-being, potentially reducing the likelihood of creating a character with “dead eyes”. 
• Designing robots and other characters to be “cute” instead of more human-like, possibly making it easier and less scary for humans to engage with them. 

Depending on the exact purpose of the technology, robot creators, computer designers, and other professionals in the technology industry may take other steps to avoid falling into the uncanny valley. 

As computer users and potential interactors with robots and other artificial characters, you may use this information to reflect on your relationship with technology and consider how these tools could affect your mental health. 

Improving your relationship with technology in therapy

Therapy can offer a space for people to explore their use of technology and its potential impact on their mental health and relationships.

Some people prefer to begin this process in person, while others prefer the ease and flexibility of a digital platform. Many clients find it easier to access therapy online because it often reduces logistical challenges commonly associated with in-office therapy.

In many cases, online therapies are just as effective as in-person therapy, as evidenced by a 2015 study of online cognitive-behavioral therapy for anxiety and depression. At the posttreatment assessment and three-month follow-up, the research team measured significant improvements in patients’ anxiety and depression, which were comparable to the outcomes of face-to-face CBT treatments cited in the study.