Hello Brain. Welcome to the Uncanny Valley

It’s the familiar colliding with the alien. Our primal instincts want to welcome the android into the pack, even while other evolutionary instincts tell us to bash its head with the nearest bone. – Erik Sofge

The “uncanny valley” is an evocative and poetic term used in the field of robotics to capture that feeling of “strangeness” when we see or interact with an almost human-looking robot. But is the uncanny valley real or is it all in our mind? And if it is real, will the uncanny valley hinder future interactions between humans and social robots?

Mori’s “Valley of Eeriness”

In a 1970 essay, robotics professor Masahiro Mori used the term “bukimi no tani”, translated more accurately as “valley of eeriness”, to describe that feeling of “the familiar colliding with the alien”, specifically as it relates to robotics. The hypothesis has been linked to Ernst Jentsch’s concept of the “uncanny” identified in his 1906 essay “On the Psychology of the Uncanny”.

Mori’s original hypothesis states that:

• As a robot’s appearance is made more human, an observer’s emotional response to the robot will become increasingly positive and empathic.

• However, there is a point beyond which the response quickly becomes that of strong revulsion, where the robot’s appearance and motion are somewhere between “barely human” and “fully human”. This is the uncanny valley.

• But, as the robot’s appearance continues to become less distinguishable from that of a human being, the observer’s emotional response becomes positive once more.

Social robots that have descended into the “valley of eeriness” include Osaka University’s CB2, Waseda University’s KOBIAN, MIT’s Nexi, and Repliee Q2:

But not everyone is convinced that the uncanny valley is real.

A Skeptic’s Perspective

Erik Sofge, contributing editor for Popular Mechanics, isn’t totally convinced of the uncanny valley. He witnessed firsthand the interaction between humans and CB2, KOBIAN, and Nexi:

In person, no one rejected the robots. No one screamed and threw chairs at them, or smiled politely and slipped out to report lingering feelings of abject horror. […] From the stiff pivot of Nexi’s body to the way the mouth on its swollen doll head flaps open as it speaks, it’s a prime example of the howling depths of the uncanny valley. And yet, when I met Nexi, and its giant blue eyes snapped to attention, and that same freakishly child-like, engorged head—really just a mask, barely concealing a tangle of motors and cables that become visible in profile—turned to me, all social distance collapsed.

When it comes to robots, Sofge claims, the uncanny valley is “a largely hypothetical chasm” and a “fleeting, cognitive glitch that has no bearing” on the future of human-robot interaction. However, there may be some evidence to support Mori’s graph and, if correct, the uncanny valley may be neither hypothetical nor fleeting, but hardwired into our brains.

A Peek Inside ‘Brains on Androids’

Curious about the uncanny valley, Ayse Saygin and her team at University of California, San Diego set out to prove a hunch: that the “action perception system” in the human brain is tuned more to human appearance or human motion. Using functional MRI, they looked at the brain activity of people viewing videos of a human-looking android (compared to videos of a human and a robot-looking robot):

According to their interpretation of the fMRI results:

…if it looks human and moves like a human, we are OK with that. If it looks like a robot and acts like a robot, we are OK with that, too; our brains have no difficulty processing the information. The trouble arises when – contrary to a lifetime of expectations – appearance and motion are at odds.

Have Saygin and her colleagues discovered evidence for the uncanny valley? If so, will social robots fail to evoke the empathic response required for productive human-robot interactions? Saygin doesn’t think so:

As human-like artificial agents become more commonplace, perhaps our perceptual systems will be re-tuned to accommodate these new social partners.

Alternatively, Mori recommended that roboticists should “create—using nonhuman designs—devices to which people can relate comfortably.”