Introducing the Uncanny Valley

I'm interested in the uncanny valley: often described as the sense of eeriness and unease that accompanies sight of something that is almost but not quite human, the ‘uncanny valley’ was once a little-known idea relating to robotics design and how people’s reactions to robots changed as they were made to look increasingly humanlike.

Some examples of things that might fall into the uncanny valley include computer game characters where the appearance and behaviour are mismatched, department store mannequins with realistic faces but blank eyes, and animated characters where the synchronized voices are just slightly mis-timed. The uncanny valley is best described with reference to the following graph:

The graph is based on an idea described by Japanese Roboticist Masahiro Mori in 1970 and describes two proposed paths for how a sense of familiarity can vary according to human likeness on a continuum from the extremely un-humanlike (an industrial robot) to completely human (a healthy living person).

The three examples that appear in this valley – the zombie, corpse and prosthetic hand – are all certainly unsettling and would cause a sense of disquiet. Although Mori did not offer additional examples, it could be seen that clowns, masks and dolls designed to be particularly realistic could also live in the same valley. The idea seems an intuitively appealing explanation for familiar experiences: as technology has advanced and it is no longer rare to encounter a highly realistic computer game or film character, the uncanny valley has become better known. Recently digital artists have begun to proclaim that their latest creation has ‘leapt’ the uncanny valley to the extent that coverage sometimes now moves from the technical domain to mainstream media. However, the uncanny valley is a descriptive rather than an explanatory idea as the graph is based on Mori’s subjective judgments of the familiarity/strangeness of each example and the curves linking them are illustrative rather than based on any empirical findings. Human likeness is presented as a percentage but the placement of the examples seems arbitrary: a stuffed animal is arguably less humanlike than a humanoid robot rather than more as it appears on the graph.

From a psychological perspective, there are two fundamental questions to be addressed: what is the uncanny valley and what causes it to occur? This can be further expanded into whether it has a cognitive and/or emotional basis, and what are the qualities of those entities that elicit it. Immediately, the uncanny valley has a range of possible explanations: does it occur because we are particularly attuned to perceiving human likeness and are therefore overly aware of where there are flaws? Is it a primal fear that we could be ‘replaced’ by something so similar to us yet so different? Is it triggered because we fear our own mortality, have a bias towards the familiar and away from strangeness, or want to avoid ‘faulty genes’? These possibilities give a tantalizing set of questions to investigate, and this website is helping me to answer those by conducting experiments online.

Stephanie Lay, 2009.


Reference:

MORI, M., 1970. The Uncanny Valley. Energy, 7(4), pp. 33-35 translated as ‘The Valley of Familiarity’ in MacDorman, K., 2005. Androids as an experimental apparatus: why is there an uncanny valley and how can we exploit it? CogSci-2005 Workshop: Toward Social Mechanisms of Android Science, July 25 – 26 2005, pp4-8.

Latest public research journal entries