The uncanny valley itself is where dwell monsters, in the classic sense of the word. Frankenstein’s creation, the undead, the ingeniously twisted demons of animé and their inspirations from legend and myth, and indeed all the walking terrors and horrors of man’s imagining belong here. In essence, they tend to be warped funhouse-mirror images of humanity, and many if not most share one or both of a pair of common traits. —Dave Bryant —The Uncanny Valley
Bryant presents the work of Masahiro Mori, who noted that people like dolls and toys that represent humans, but that as these items start to look and behave more human, there is a sudden drop in the graph. [Correcting that last bit, which got cut off.]
Now, I’d have to look into Mori’s work more closely to determine exactly what he was measuring, or why he chose to see a connection between these objects:
- industrial robot (slightly positive)
android (more positive)
moving corpse (most negative)
prosthetic hand (somewhat negative)
handicapped person (neutral)
bunraku puppet (most positive so far)
unhealthy person (higher than the previous one)
healthy person (even higher)
When you list them in that order, you get a positive “emotional response” curve from the industrial robot to the andriod, and then you get the “uncanny valley” — that is, the huge drop in the graph — for “moving corpse,” followed by an upward sweep towards “healthy person.”
But what is it about a moving corpse that necessitates that it should be placed to the right of an android on the above scale? The “valley” disappears if you simply sort the items in a different order, like this:
- moving corpse (lowest)
prosthetic hand
handicapped person (netural)
industrial robot
android
bunraku puppet
unhealthy person
healthy person (highest)
My point is not that Bryant is somehow being dishonest, or that Mori’s work was somehow unscientific. But the choice to arrange objects in order to create a visual “ucanny valley” in the graph of emotional responses is a rhetorical device, designed to draw attention to a particular measurement and to control the context in which that measurement is presented.
Um, “there is a sudden drop in the “…..?
“how is a corpse (which is certainly the shape of a human) less anthropomorphic than a prosthetic hand”
I actually agree with you on this point, but it turns out that it doesn’t really matter if the prosthetic hand and the corpse switched positions according to which one was more anthropomorphic – they’re all 3 in the valley anyhow. (see wikipedia graphs pulled from original article)
I like this theory, and I use it in my classes on virtuality, etc because it’s something the students can grab onto immediately and begin debating. We especially use it in discussions of Blade Runner (film) and Second Life (virtual world).
I agree that the sorting order isn’t random, but how is a corpse (which is certainly the shape of a human) less anthropomorphic than a prosthetic hand? Thanks for pointing out that link — we’re doing a unit on science and statistics in news writing, so that will be worth a look.
I love all things uncanny, and found this really interesting.
You asked, “But what is it about a moving corpse that necessitates that it should be placed to the right of an android on the above scale?” I think it’s ordered according to anthropomorphism — that the response is more negative the closer the simulacrum approaches humanity, but there’s clearly some degree of subjectivity still evident in all of this. Still, I think there’s some merit to this, though, in terms of phenomenological theory. There’s more coverage on this fascinating topic in the Wikipedia at http://en.wikipedia.org/wiki/Uncanny_Valley — worth a look-see, since others claim it’s pseudoscience.