Select Page

The Uncanny Valley

What Have I Been Reading?

Subscribe via Email

If you enjoy my ramblings, then you can sign up using your email address and get notified whenever I post!

Join 372 other subscribers

by | Jan 25, 2019

Following on from my previous post, on What Makes a Robot? , I stumbled upon a phenomenon known as “The Uncanny Valley”.

Reading Asimov’s “I, Robot” I find myself sympathising with many of the robots in the stories: when they are unfairly accused or used as unpaid labour with little rights.  In Asimov’s book, all of the newer robots have “positronic brains” which allow them to have personalities, rather than just being number crunching machines. On reflection, I assume this greatly influenced my feelings towards them at certain points of the book. However, when I imagine the robots I have an image similar to WALLE or Baymax.

Image result for Baymax

I find that, instead, picturing a robot similar to Hiroshi Ishiguro’s robot (Geminoid) gives me the creeps and I very quickly feel less sympathetic. Even if they do have a lovely personality!

This would be described as falling into the “Uncanny Valley”. The term was coined in 1970 by a robotics professor Masahiro Mori.

Masahiro Mori found that human beings don’t relate or feel anything for robots that have no human similarities, like a robot hoover or a robotic arm. What we do love, are robots with human like features: a robot that has human form, human-like eyes, detailed facial features. Importantly, though it remains obviously a robot (made of metal, no skin; the usual tell tale signs). We find it “cute”, this imitation of human beings. This positive response continues to increase as the robots become closer and closer to resembling a human copy BUT there is a point where the robot is too human looking but not human-like enough. It looks human, but something isn’t quite right and we [humans] are unsettled, uneasy, and no longer feel familiarity nor empathy towards these humanoids. This is the uncanny valley. When something appears human, but it doesn’t act or move in the same ways that we do. Maybe the eyes are just a little too sparkly, it’s movements janky or it blinks in an awkward way.

Interestingly, this is an issue in games and CGI films, not just robotics. Here is the graph:

You can see zombies feature: human like, but just not right! [I am not a zombie film fan for this exact reason]. You will notice though that after this dip into the uncanny valley, there is an upward trend. We have no problem with something that looks, sounds and moves like it is “supposed to”. Therefore, at present, films, robotocists and game developers are attempting to either achieve the point just before the dip of the “uncanny valley” or bridge it and reach the upward trend on the other side. As Thalia Wheatley, psychologist at Dartmouth College, points out “Pixar took a lesson from ‘Tin Toy, we have to nail the human form or not even go there.”

There is a fantastic youtube video discussing this “Video Games and the Uncanny Valley”.

At the point of writing, there is no agreed upon scientific/psychological explanation for the “Uncanny Valley”. Some suggestions include:

  • Explanation One: When we perceive something as human, our brain makes certain assumptions about how it should behave/move/act and when it doesn’t our brains have to re-evaluate and this causes the uneasy feeling. Sort of like if you went up to a store assistant to ask where the cute tops are and they stray from the expected response and yell donkey noises at you.
  • Explanation Two:Human beings are very protective over our minds. For example, we like to think it is a characteristically human-trait to have a personality, be able to hold conversations and respond to other humans emotions. When machines become more than number crunchers or an “act” we become threatened and feel unsettled.

Some further explanations are summarised in the following article “What Makes the Uncanny Valley so Unsettling?“. Despite having an explanation for why I may feel unsettled, Geminoid still gives me the heeby jeebies.

I feel less unsettled by Sophia because (explanation one) although she has human-like features and her face does resemble a human being, she remains very obviously a robot. Hanson Robotics have left behind her head exposed to show the inner workings, I believe this may be in an attempt to dodge the uncanny valley. Explanation two, I have read many articles and [at the time of writing] Sophia appears to be a highly sophisticated chatbot rather than a free-thinking machine. Her answers are pre-programmed, those witty on liners aren’t made up on the spot. Therefore, her mind remains obviously a machine that can be programmed.

How do you all feel? Do you have thoughts or strong feelings towards robots that appear human?