The Uncanny Valley

I’ve been thinking about the Uncanny Valley. In case you’re not familiar with it, see:

I think the graph should be more complex than it is. I think there are more things at work. I think the idea actually represents several different variables, such as Emotional Connectivity and Abstract/Human quality.

I also think it actually has less to do with the creatures we’re seeing and more to do with our own reflections about the predictability and trustworthiness of the world.

The dip appears to be a dip because it represents approaching a threshold of trust.

Hypothesis: People consider other people inherently untrustworthy.

Rocks, robots, machines, and obvious items are recognized as the rules-based mechanisms they are. This includes animations, stuffed animals, etc. We instinctively understand that no matter how “real” our interaction with a stuffed animal is, for example, it’s always a stuffed animal and will always do as we expect (or as we role-play – whatever).

The Uncanny Valley represents a threshold beyond which we start considering the object sufficiently human to earn the label of untrustworthiness.

It is vital to understand that the Uncanny Valley indicates a transition, a place where something once completely trustworthy (or rather, “predictable”) becomes untrustworthy (or “human”). Only in the past few decades has this even been close to possible.

An object must shift for it to encounter the Uncanny Valley effect. In other words, a wax figure will not particularly creep us out unless it moves, because we know it’s a wax figure. Our categorization of it is static. Once it moves, and if it moves realistically, we must on-the-fly recategorize it, and if that recategorizing action brings it closer to being human, then we question our trust, and thus it appears creepy.

Examples of things that experience recategorization making them more creepy: dolls that come alive. Software that uses algorithms to talk with people (I’m looking at you, Eliza!).

Examples of things that experience recategorization making them less creepy (indicated by relief): things we discover are just machines, are just dead, or are just inanimate (such as piles of coats).

I think as we mature technologically, we will adjust our emotional expectations of such things as machines. I think as we mature technologically, we will find the notion of robots being “creepy” as antiquated and hilarious as phlogiston and lizard people in the center of the Earth.