Total Recoil: The Uncanny Valley is an Uncanny Cliff
Every act of perception involves an act of categorization
This is part two of a series on what ‘uncanny’ and the ‘uncanny valley’ mean and how to accurately use them to describe experiences (in VR or otherwise).
TL;DR
- The uncanny valley is specifically about the relationship between being human-like and likability.
- Categorization is part of perception.
- Science shows that toy non-humanoid robots (think WALL-E) are preferred to human-like ones (and actual humans!)
Background
1919 — Sigmund Freud published an essay called The Uncanny” (translated from the German unheimlich), defining it to mean weird/eerie/unfamiliar.
1970 — Valley of Eeriness (translated from the Japanese bukimi no tani) was coined by robotics professor Masahiro Mori to model affinity for androids as they become more humanlike.
1978 — Uncanny Valley first appeared in English inside a book by Jasia Reichardt called “Robots: Fact, Fiction, and Prediction.”
2007 — Uncanny Cliff was introduced to more accurately reflect the shape of the curve.
What Is the Uncanny Valley?
The uncanny valley is a very precise feeling of weirdness or fear that nearly-humans (but not quite humans) evoke. The uncanny valley is elicited when a robot or AI has some human likeness, but isn’t human. Long time art director and VR/AR insider Spencer Lindsay uses “creepy corpse” in his definition of the uncanny valley.
You can see the valley in the graph below:
It’s assumed that liking of a robot increases the more human like that it is, until it crosses into the valley (the grey area). Lindsay’s creepy corpse would appear to be at the nadir of the valley — with zombies!
It is likely that the uncanny valley response is evolutionarily adaptive. Studies done with rhesus monkeys show a similar pattern of likability. Monkeys will look longer at the unrealistic (left) and real (right) faces below, rather than the realistic one in the center.
The Uncanny Valley Confuses Categories.
To categorize is a fundamental psychological process that happens automatically. Every act of perception involves an act of categorization. You haven’t completed the process of perception until you have categorized it and matched it up to other things that you know about. That is, “this object is the same as other things I know about and different from these other things.” We judge new things based on their similarity to previous things. And a human-like puppet easily confuses our judgment process.
“It’s the process that grinds away constantly and generates much of our understanding and response to the world,” says Berkeley psychologist Dacher Keltner. “First of all, it’s how do you categorize things? And that’s everything. Do I sleep with him or not? Is that a boy or a girl? Is that predator or prey? If you solve how this process works, then you solve how you know things. It’s how knowledge about the world is organized. It’s like the thread that is woven through everything in the mind.”
“Do I sleep with him or not? Is that a boy or a girl? Is that predator or prey? If you solve how this process works, then you solve how you know things.”
Encountering a creepy android from the trough of the uncanny valley leads to conflicting perceptual cues. Is it human or not? It interferes with the automatic, System 1 processing that we rely on to get through our day.
The Uncanny Valley Is an Uncanny Cliff
Lastly, there have been empirical studies that map the uncanny valley since Mori originally theorized it. Researchers show 11 different images as an object morphs from a thing to a human, and there is a decrease in liking midway through.
It’s actually more like a cliff, because likability doesn’t fully recover:
“The uncanny valley appears to be more of a cliff than a valley since even pictures of humans do not reach the level [of likability] of pictures of toy robots. It has to be acknowledged that there is a small upwards trend again towards highly human-like entities, which results in a small valley. However, the most dominant feature in the graph is not the valley, but the cliff preceding it.” — Bartneck et al.
Takeaways for Designers:
- Avoid anything related to the uncanny valley by populating your experience with non-humanoid avatars. People prefer non-humanoids to humanoids
- The uncanny valley doesn’t actually exist. It’s more like a cliff. Once you get too humanoid, liking decreases dramatically and never fully returns.
Additional Info:
If you are interested in the uncanny valley, you’ll like Kimberley Voll’s interview on the fidelity contract. Listen to her full interview with @VoicesofVRKentBye here.
Bartneck, C., Kanda, T., Ishiguro, H., & Hagita, N. (2007). Is the Uncanny Valley an Uncanny Cliff? Proceedings of the 16 th IEEE International Symposium on Robot and Human Interactive Communication, RO-MAN 2007, Jeju, Korea, pp. 368–373.
H/T to Spencer Lindsay for letting me quote him here.
This post originally appeared on The Extended Mind blog
