Why almost-human is creepier than clearly not human
In 1970, Japanese roboticist Masahiro Mori proposed a startling hypothesis: as robots become more human-like, our affinity for them increases—but only up to a point. Just before reaching full human likeness, there's a dramatic plunge into revulsion. He called this dip bukimi no tani—the "uncanny valley."
The Core Paradox: A crude, obviously non-human robot feels friendly and approachable. But a robot that's 95% human-like triggers deep unease—it's "almost right" yet somehow deeply wrong. Cross the valley completely to 100% human, and comfort returns.
This valley isn't just about robots. It explains why certain CGI characters, wax figures, realistic dolls, and even corpses evoke such visceral discomfort. The closer something gets to human without quite making it, the more disturbing it becomes.
Mori's original graph plots human likeness (horizontal axis) against affinity/comfort (vertical axis). The curve rises steadily—until it suddenly drops into the valley of eeriness.
Movement intensifies the effect. Mori noted that a moving zombie is far more unsettling than a still corpse. The red curve shows how movement creates both higher peaks and deeper valleys—making the uncanny effect even more pronounced.
Drag the slider to traverse from cartoon to human. Notice how your comfort level changes—particularly in the middle zone where things feel "wrong."
The uncanny valley has haunted filmmakers, game developers, and roboticists. These examples show what happens when creations fall into—or narrowly avoid—the valley.
The Pixar Strategy: After their 1988 short "Tin Toy" provoked negative reactions to its realistic baby, Pixar deliberately moved away from photorealism. Their stylized characters feel more "alive" than hyper-realistic CGI humans.
Mori specifically noted that movement intensifies the uncanny effect. A still mannequin is unsettling; a moving one is terrifying. Watch how the same face feels different when animated:
The third face moves almost naturally—but something's subtly wrong. This triggers the uncanny response.
Scientists have proposed several explanations for the uncanny valley response:
Imperfections in near-human faces may trigger disease-detection mechanisms. Evolution programmed us to avoid sick individuals—and something "almost human" might signal illness or genetic abnormality.
Near-human entities can remind us of corpses or dying individuals. Mori himself noted this: something that looks alive but isn't—or looks dead but moves—evokes existential discomfort.
Our brains struggle to categorize near-human entities. Is it human or not? This cognitive conflict—the amygdala detecting threat while the visual cortex says "human"—creates discomfort.
When something looks human, we expect human behavior. Subtle deviations—a too-smooth blink, a slightly wrong smile—violate these expectations and feel deeply wrong.
Research using fMRI scans has traced the uncanny valley to specific brain regions. The amygdala—linked to fear and emotional processing—shows heightened activity when subjects view near-human figures. Meanwhile, visual cortex regions that track "human-likeness" create conflicting signals, producing the cognitive dissonance we experience as eeriness.
Masahiro Mori publishes "Bukimi no Tani" in the obscure Japanese journal Energy. It receives almost no attention for decades.
British art critic Jasia Reichardt translates "bukimi no tani" as "uncanny valley"—giving the concept its now-famous English name.
Pixar's "Tin Toy" features a realistic baby that disturbs audiences—the film industry's first major encounter with the valley.
"Final Fantasy: The Spirits Within" attempts photorealistic humans. Despite technical achievement, characters feel "dead" and the film flops.
"The Polar Express" becomes the poster child for uncanny CGI. Reviews call it "creepy" and "horrifying."
Mori's essay is finally translated to English. Research explodes across robotics, film, and cognitive science.
"Cats" becomes a cultural phenomenon for all the wrong reasons. Sonic is redesigned after trailer backlash.
Masahiro Mori passes away at age 97. AI-generated faces approach—but haven't yet crossed—the valley.
Mori's original essay was based on intuition and observation, not scientific experiment. He described watching a robotic hand at the 1970 World Exposition in Osaka and feeling a flash of eeriness when its movements seemed almost—but not quite—natural.
Since 2005, researchers have sought to prove and measure the uncanny valley:
Mori's recommendation: robot designers should not aim for full human likeness. Instead, stay safely on the near side of the valley—like Pixar characters, WALL-E, or R2-D2—or leap completely across it. The middle ground is dangerous territory.
Key Insight: The uncanny valley reveals something profound about human perception: we're exquisitely tuned to detect subtle wrongness in human faces and movements. This sensitivity—perhaps evolved for disease detection or identifying "in-group" members—now haunts our attempts to create artificial humans.
Mori, M. (1970). Bukimi no tani [The uncanny valley]. Energy, 7(4), 33-35. Translated by MacDorman, K. F., & Kageki, N. (2012). IEEE Robotics & Automation Magazine, 19(2), 98-100.