CGI and the Uncanny valley

Audrey Hepburn, dead these 21 years, digitally dug up from the grave and reanimated to shill for Dove chocolates. You may already have seen this commercial, which began running on Oscars night and is now moving into wider rotation.

Dove and the commercial producers are inordinately proud of their achievement. According to their publicity handout, they first tried hiring an Audrey Hepburn double, then tweaking her features with computer-generated imagery, or CGI. That wasn't good enough, so they decided to go all-CGI. The digital effects firm Framestore "set about building a facial rig using the FACS [facial action coding system] head scans as reference for the multitude of shapes the human face can achieve. For each of these shapes, combinations had to be carefully created to allow the face to blend convincingly between expressions during the animation process, which was pivotal to authentically reconstructing the actress’s face."

The final result is impressively almost-real but suffers from what digital effects experts know as "the uncanny valley" -- when an image is just close enough to resemble real life that it's offputting. Public reaction to the commercial seems to range from delight at seeing Audrey prancing about again on the Italian Riviera to horror at the violence done to her memory.

A number of films that use computer-generated imagery to show characters have been described by reviewers as giving a feeling of revulsion or "creepiness" as a result of the characters looking too realistic.  Examples include:

    According to roboticist Dario Floreano, the animated baby in Pixar's groundbreaking 1988 short film Tin Toy provoked negative audience reactions, which first led the film industry to take the concept of the uncanny valley seriously.

    Several reviewers of the 2004 animated film The Polar Express called its animation eerie.  CNN.com reviewer Paul Clinton wrote, "Those human characters in the film come across as downright... well, creepy.  So The Polar Express is at best disconcerting, and at worst, a wee bit horrifying."  The term "eerie" was used by reviewers Kurt Loderand Manohla Dargis, among others. Newsday reviewer John Anderson called the film's characters "creepy" and "dead-eyed", and wrote that "The Polar Express is a zombie train."   Animation director Ward Jenkins wrote an online analysis describing how changes to the Polar Express characters' appearance, especially to their eyes and eyebrows, could have avoided what he considered a feeling of deadness in their faces.

    In a review of the 2007 animated film Beowulf, New York Times technology writer David Gallagher wrote that the film failed the uncanny valley test, stating that the film's villain, the monster Grendel, was "only slightly scarier" than the "closeups of our hero Beowulf’s face... allowing viewers to admire every hair in his 3-D digital stubble."

    In the 2010 film The Last Airbender, the character Appa, the flying bison, has been called "uncanny".  Geekosystem's Susana Polo found the character "really quite creepy", noting "that prey animals (like bison) have eyes on the sides of their heads, and so moving them to the front without changing rest of the facial structure tips us right into the uncanny valley".

    The 2011 film Mars Needs Moms, the second biggest box office bomb in history, was widely criticized for being creepy and unnatural because of its style of animation.  Although being a Disney film, this movie's poor box office turnout was likely based on the negative word of mouth caused by moviegoers unsettled by the characters' weird and inhuman look.

By contrast, at least one film, the 2011 The Adventures of Tintin, was praised by reviewers for avoiding the uncanny valley despite its animated characters' realism.  Critic Dana Stevens wrote, "With the possible exception of the title character, the animated cast of Tintin narrowly escapes entrapment in the so-called 'uncanny valley.'"  Wired Magazine editor Kevin Kelly wrote of the film, "we have passed beyond the uncanny valley into the plains of hyperreality."


the uncanny valley is a hypothesis in the field of human aesthetics which holds that when human features look and move almost, but not exactly, like natural human beings, it causes a response of revulsion among some human observers. The "valley" refers to the dip in a graph of the comfort level of humans as subjects move toward a healthy, natural human likeness described in a function of a subject's aesthetic acceptability. Examples can be found in the fields of robotics[1] and 3D computer animation, among others.

Mori's original hypothesis states that as the appearance of a robot is made more human, some human observers' emotional response to the robot will become increasingly positive and empathic, until a point is reached beyond which the response quickly becomes that of strong revulsion. However, as the robot's appearance continues to become less distinguishable from that of a human being, the emotional response becomes positive once again and approaches human-to-human empathy levels.

This area of repulsive response aroused by a robot with appearance and motion between a "barely human" and "fully human" entity is called the uncanny valley. The name captures the idea that an almost human-looking robot will seem overly "strange" to some human beings, will produce a feeling of uncanniness, and will thus fail to evoke the empathic response required for productive human-robot interaction.

A number of theories have been proposed to explain the cognitive mechanism underlying the phenomenon:

    Mate selection. Automatic, stimulus-driven appraisals of uncanny stimuli elicit aversion by activating an evolved cognitive mechanism for the avoidance of selecting mates with low fertility, poor hormonal health, or ineffective immune systems based on visible features of the face and body that are predictive of those traits.

    Mortality salience. Viewing an "uncanny" robot elicits an innate fear of death and culturally-supported defenses for coping with death’s inevitability.... [P]artially disassembled androids...play on subconscious fears of reduction, replacement, and annihilation:  A mechanism with a human facade and a mechanical interior plays on our subconscious fear that we are all just soulless machines.  Androids in various states of mutilation, decapitation, or disassembly are reminiscent of a battlefield after a conflict and, as such, serve as a reminder of our mortality.  Since most androids are copies of actual people, they are doppelgängers and may elicit a fear of being replaced, on the job, in a relationship, and so on.  The jerkiness of an android’s movements could be unsettling because it elicits a fear of losing bodily control."

    Pathogen avoidance. Uncanny stimuli may activate a cognitive mechanism that originally evolved to motivate the avoidance of potential sources of pathogens by eliciting a disgust response. "The more human an organism looks, the stronger the aversion to its defects, because  defects indicate disease,  more human-looking organisms are more closely related to human beings genetically, and the probability of contracting disease-causing bacteria, viruses, and other parasites increases with genetic similarity." Thus, the visual anomalies of android robots and animated human characters have the same effect as those of corpses and visibly diseased individuals: the elicitation of alarm and revulsion.[citation needed]

    Sorites paradoxes. Stimuli with human and nonhuman traits undermine our sense of human identity by linking qualitatively different categories, human and nonhuman, by a quantitative metric, degree of human likeness.

    Violation of human norms. The uncanny valley may "be symptomatic of entities that elicit a model of a human other but do not measure up to it". If an entity looks sufficiently nonhuman, its human characteristics will be noticeable, generating empathy. However, if the entity looks almost human, it will elicit our model of a human other and its detailed normative expectations. The nonhuman characteristics will be noticeable, giving the human viewer a sense of strangeness. In other words, a robot stuck inside the uncanny valley is no longer being judged by the standards of a robot doing a passable job at pretending to be human, but is instead being judged by the standards of a human doing a terrible job at acting like a normal person. This has been linked to perceptual uncertainty and the theory of predictive coding.

    Religious definition of human identity. The existence of artificial but humanlike entities is viewed by some as a threat to the concept of human identity, as constructed in the West and the Middle East. This is particularly the case with the Abrahamic religions (Christianity, Judaism, and Islam), which emphasize human uniqueness. An example can be found in the theoretical framework of psychiatrist Irvin Yalom. Yalom explains that humans construct psychological defenses in order to avoid existential anxiety stemming from death. One of these defenses is "specialness", the irrational belief that aging and death as central premises of life apply to all others but oneself. The experience of the very humanlike "living" robot can be so rich and compelling that it challenges humans' notions of "specialness" and existential defenses, eliciting existential anxiety. The creation of human-like, but soulless, beings is considered unwise; the golem in Judaism is a well-known example. Like anthropomorphic robots, a golem may be created with good intentions, but its absence of human empathy and spirit can lead to disaster.

    Conflicting perceptual cues. It is hypothesized that the uncanny valley effect is associated with a more general psychological phenomenon in which negative affect is produced by the activation of conflicting cognitive representations. Perceptual tension occurs when there are conflicting cues to category membership, such as when a humanoid figure has robotic movement, which is then experienced as psychological discomfort (i.e., "eeriness"). This is supported by converging lines of evidence, including behavioural and neuroimaging studies, and mathematical modelling. First, Burleigh and colleagues demonstrated that faces at the midpoint between human and non-human categories produced a level of reported eeriness that diverged from an otherwise linear model relating human-likeness to affect. Second, Saygin et al. found increased neural activity in the parietal cortex when participants were viewing a category-inconsistent robot (i.e., a robot with a human-like appearance and robotic movement), as compared to when they were viewing a category-consistent robot (i.e., a robot with a robot-like appearance and robotic movement) or a human, and suggested that this activity indicated increased prediction error due to perceptual conflict.Moore used a Bayesian mathematical model to provide a quantitative account of perceptual conflict.This model not only predicts the shape of Mori's hypothesized curves, but may also allow predictions to be made for a range of social situations in which conflicting perceptual cues might give rise to negative reactions.

wiki