The uncanny valley – good or bad?

Who didn’t feel a general sense of warmth from Alyx Vance’s smile in Half-Life 2, read the exhaustion off the tortured Old Snake in Metal Gear Solid 4, or chuckle at Nathan Drake’s sly smirk in the face of Uncharted’s ruthless pirates? The representation of the human likeness in video games has been inching towards photo-realism in the recent years as polygon counts have exploded and motion capturing has become more expressive.

Games featuring technological eye-candy are often rewarded with undivided attention from consumers, bedazzled by how “realistic” games can truly be. However, while the multiple cell processors of game platforms and the increased production times of future releases threaten to erase the perceptual boundaries between Man and their virtual creations, there is one possible obstacle land-mining the front entrance to this “Eden” of simulated humanity.

In 1970 the Japanese roboticist Masahiro Mori wrote an essay discussing an effect that increasingly human-looking robots would have a on observers called the “Uncanny Valley”. He observed that as rudimentary looking robots began to take on more and more human characteristics, observers would find them more and more relatable.

Yet, at the point where these robots became almost identical to human beings with synthetic skin and increased body animations, this attraction would plummet into disgust and revulsion as the robot’s disturbing inaccuracies became more noticeable.

This trend would continue until the robot took the final steps towards becoming completely human-like and the details which originally made the robot repulsive disappeared. At this point the observers’ attraction would shoot back up and their revulsion would decrease.

On a line graph comparing the familiarity of the observers to the humanness of the robots, this ironical trend would create what looked like a valley, a result open to a few different explanations.

As human beings ourselves we know what another human being is supposed to look and act like. Anything which appears “off” in something being judged within these standards of humanity comes across as unsettling. The effect could also come from the parallels that the “off” qualities in a human-like robot make with the features of seriously ill individuals and corpses, people who naturally elicit alarm and disgust in their observers.

The Uncanny Valley could even be an instinctual carry-over that allowed ancestors to avoid inbreeding with other intelligent primates. Yet, whatever the cause, this presents a problem for video games as it makes extremely difficult the jobs of any game creators looking to achieve photo-realism with their characters.

Any discrepancy from normal human characteristics and their attempt at perfection plummets down into the valley, turning their creations from relatable protagonists to disquieting malformed creatures.

So what could these game makers do? Well, while some creators feel that photo-realism should still be striven for in order to advance the industry as a whole, they also feel the focus should be placed on the peak located before the chart’s Uncanny Valley instead of the one after, stylization instead of realism.

Here characters which are clearly not human but rather synthetics with human-like qualities (an example would be the exaggerated features found in a character like Mario) avoid the pit-falls and linearity found in realism and instead open up the endless possibilities found in stylization.

Essentially, there’s only one way to make a “real” person look real but there are millions of ways to stylize something not meant to be real. If game creators allow the human qualities to stand out in a mostly inhuman character instead of the opposite, then maybe the Uncanny Valley can be avoided.