What Makes the Uncanny Valley So Unsettling?
Every day we inch closer to the time when intelligent robots will be a part of everyday life. Among the many challenges we must overcome before then is gaining a better understanding of the “uncanny valley” — the feeling of discomfort we have around humanlike robots. Thus far, most research has tended to focus on robot appearances. For example, there is evidence that humanlike robots are unnerving because their faces remind us of death, have abnormal features, and fail to align with our expectations. Still, the research has been relatively inconclusive and it’s failed to uncover a more comprehensive explanation.
Kurt Gray and Daniel Wegner attempted to remedy this problem by examining whether a robot’s potential to experience (i.e. to feel and sense) is what drives a person’s discomfort rather than a robot’s agency (i.e. the ability to act and do.) First, Gray and Wegner demonstrated that humanlike robots elicit greater ascriptions of experience and feelings of discomfort. In a second experiment, they found that participants were more unnerved by a machine that could experience emotions than a machine that was capable of taking actions. In a third and final experiment, participants were more unsettled by a mentally disabled person who couldn’t experience emotions than they were by a mentally disabled person who couldn’t perform normal actions. If Gray and Wegner’s results are supported and extended in future work, the idea that the uncanny valley is tied to perceptions about the capacity to experience could provide the kind of broad explanation that’s been missing.
Gray and Wegner also believe their findings can teach us something about our perception of what it means to be human:
Although this research focused on strange minds—on feeling robots and unfeeling people—it speaks to the broader idea about what makes us human. Higher cognition may separate us from animals and feature prominently in explicit definitions of the human mind (Aristotle, BC350), but feelings of unease indicate that experience is implicitly viewed as more essential to humans.
The big question is what it is about the ability to experience that’s so powerful. I think one answer may lie in the fact that people are generally good and don’t like hurting others for no reason. When it comes to the feelings of other humans, we’re good at predicting the consequences of our actions. But with robots we have no idea how or why they feel certain things, and the lack of knowledge about what might causes a robot to experience something negative makes us uncomfortable. In other words, robots don’t creep us out because they can harm us, they creep us out because we might unintentionally harm them. Interacting with a robot that seems capable of experiencing might be like an extreme version of culture shock. You’re uncomfortable because you have no idea how your actions will be interpreted and you’re afraid you’ll do something terrible.
Gray K., & Wegner D.M. (2012). Feeling robots and human zombies: Mind perception and the uncanny valley. Cognition PMID: 22784682