What Makes the “Uncanny Valley” So Unsettling?

Scientific explanations for this unusual phenomenon

Creepy Life-like Dolls
Carol Yepes / Getty Images.

Have you ever looked at a life-like doll and felt your skin crawl? Gotten an unsettled feeling when you saw a human-like robot? Felt nauseous while watching an on-screen zombie lumber around aimlessly? If so, you’ve experienced the phenomenon known as the uncanny valley.

First proposed in 1970 by Japanese roboticist Masahiro Mori, the uncanny valley is the creepy, repulsed feeling we get when we observe an entity that looks almost human, but lacks some essential element of humanity.

Characteristics of the Uncanny Valley

When Mori first proposed the phenomenon of uncanny valley, he created a graph to explain the concept:

Mori's Uncanny Valley Graph translated by MacDornan and Minato
Mori's Uncanny Valley Graph translated by MacDornan and Minato.  Wikimedia Commons

According to Mori, the more "human" a robot appears, the more positive our feelings towards them will be—up to a point. As robots approach near-perfect human likeness, our responses quickly turn from positive to negative. This sharp emotional dip, seen in the graph above, is the uncanny valley. Negative responses can range from mild discomfort to severe repulsion.

Mori’s original graph specified two distinct pathways to the uncanny valley: one for still entities, like corpses, and one for moving entities, like zombies. Mori predicted that the uncanny valley was steeper for moving entities.

Finally, the uncanny valley effect subsides and people’s feelings towards a robot again turn positive once the robot becomes indistinguishable from a human being.

In addition to robots, the uncanny valley can apply to things like CGI movie or video game characters (such as those from The Polar Express) whose appearance doesn’t match their behavior, as well as wax figures and realistic-looking dolls whose faces look human but lack life in their eyes.

Why the Uncanny Valley Freaks Us Out

Since Mori first coined the term, the uncanny valley has been researched by everyone from roboticists to philosophers to psychologists. But it wasn’t until 2005, when Mori’s original paper was translated from Japanese into English, that research on the subject really took off.

Despite the intuitive familiarity of the idea of the uncanny valley (anyone who’s ever seen a horror movie featuring a human-like doll or zombie has likely experienced it), Mori’s idea was a prediction, not the result of scientific research. Therefore, today, scholars disagree about why we experience the phenomenon and whether it even exists at all.

Stephanie Lay, an uncanny valley researcher, says she’s counted at least seven explanations for the phenomenon in the scientific literature, but there are three that show the most potential.

Boundaries Between Categories

First, categorical boundaries may be responsible. In the case of the uncanny valley, this is the boundary at which an entity moves between non-human and human. For example, researchers Christine Looser and Thalia Wheatley found that when they presented a series of manipulated images created from human and mannequin faces to participants, participants consistently perceived the images as life-like at the point where they crossed to the more human end of the spectrum. The perception of life was based on the eyes more than other parts of the face.

Perception of Mind

Second, the uncanny valley could depend on people’s belief that entities with human-like features possess a human-like mind. In a series of experiments, Kurt Gray and Daniel Wegner found that machines became unsettling when people attributed the capacity to feel and sense to them, but not when people’s only expectation of the machine was the ability to act. The researchers proposed this is because people believe the ability to feel and sense is fundamental to humans, but not machines.

Mismatch Between Appearance and Behavior

Finally, the uncanny valley may be the result of a mismatch between the appearance of a near-human entity and its behavior. For instance, in one study, Angela Tinwell and her colleagues discovered that a human-like virtual entity was regarded as most unnerving when it didn’t react to a scream with a visible startled response in the eye region. Participants perceived an entity who exhibited this behavior as having psychopathic traits, pointing to a possible psychological explanation for the uncanny valley.

The Future of the Uncanny Valley

As androids become further integrated into our lives to aid us in a variety of capacities, we must like and trust them in order for us to have the best interactions. For example, recent research suggests that when medical students train with simulators that look and behave like humans, they perform better in real emergency situations. Figuring out how to transcend the uncanny valley is critical as we rely more and more on technology to assist us in everyday life.

Sources