The Uncanny Valley might not actually exist




Keeping in mind that I have not yet read the paper, it's seems most of the research dealt with faces. I wonder if it's more of a locomotion or body movement issue?


I honestly don't get what the fuss is about.

"The Uncanny Valley" is simply a handy way of refering to a specific phenomenon. The situation it describes is real - or at least the basic facets of it are. Certain kinds of human-looking objects invoke certain kinds of "creep-out" effects in certain people. Referring to that response as "The Uncanny Valley" seems entirely reasonable as just a useable label.

Mori wasn't a scientist performing research on human behavior - he was a roboticist describing an effect that he had seen in action. People were, and still are, creeped out by certain human-like things, and he, as a roboticist, came into direct contact with that human response. It wasn't hypothetical - he had people telling him things like "That's creepy!" in regards to human-like robots. So it's not at all shocking that his explanations of the phenomenon are not scientifically accurate or comprehensive - he never intended for them to be!

Moreover, the BBC article is absolute rubbish. It cites the Hanson study, then immediately points out that this particular study has been criticized. So a single study that they admit is almost certainly flawed is grounds for a misleading headline about how "The Uncanny Valley" might not even exist? Talk about your senseless invent-a-controversy journalism!

The rest of the article talks about how other studies have had difficulty "mapping" the effect, feebly attempting to use that to prop up the article's earlier suppositions. Honestly? They're seriously suggesting that the fact that science hasn't yet been able to properly measure the phenomenon in easily understandable ways means the effect itself may not actually exist? Rather than the possiblity that - just maybe! - it's just one of those weird facets of our own immensely complicated human nature that is beyond our understanding for the time being?

Absolute rubbish. The BBC should be ashamed for putting out that article, and BoingBoing should be ashamed for championing it.


What? it gave you a chance to explain all that. Isn't that the point?


If you're suggesting that journalism no longer serves to inform the public but merely to entertain, then I suppose you are correct to some degree.

In you're instead trying to mock my verbose commentary on the articles, that's another matter.

It's hard to tell which. wink


I think the inherent difficulty is that "the Uncanny Valley" describes a subjective human experience. You might as well be saying: "I feel happy", "I feel pain", so it's problematic for anyone to say, "Well, what you're feeling doesn't actually exist". I get the sense that what the study/article are perhaps getting at is that the "shape" may not be what we thought; that the "valley" is not a conveniently predictable dip on an X/Y graph. So a robot can seem uncomfortably uncanny, but that doesn't necessarily mean that there's a "valley".


I have to admit I never actually thought it was a scientific concept. I was amazed how during the 1980s and on it became an increasingly prevalent term, but it is a useful term ; if you think about it in terms like Wittgenstein's language-game theory, it helps describe a feature of a world that people experience around simulated humans. Some people were experiencing something and this term helped them communicate an issue. Terms like 'uncanny' and 'sublime' are notoriously difficult to define - and I think this is going to affect any sociology done in the area. I think they are more aesthetic terms, and communicate a possible category of experience, and so will probably be constantly evolving socially. What 'uncanny valley' meant in the early 1970s has to be different from what it means today, due to changing contexts.


I honestly think things are posted on BB just to air them so to speak. No endorsement is implied. Especially the short summary things. It is like a friend saying, "Hey, did you see this." I was interested to read your response, though I thought accusing BB of unprofessionalism was over-board. My response was a bit of a tease, yes, but also my true feelings. This is why I read the comments. To find stuff out. So thanks for that.


In exacting, scientific terms? Sure, maybe there isn't a specific "dip" or "valley" that you can chart out. It might not be an exact thing.

I think the problem is the "valley" usage. Maybe Mori was wrong about it being a specific "dip" as you get closer to mirror-image human appearance and behavior. Maybe we'd be better off using other terminology to refer to the general concept of "The Uncanny".

But one important thing to note is that Mori was writing with a characteristic Japanese style, and that he was making a sort of poetic comparison between the goal of designing robots that people found appealing and the effort of climbing mountains. Thus, in constructing that particular metaphor, the reference to valleys seems to have come along completely naturally.


No endorsement implied? Really?

Eveleth's piece is definitely worth a read.

That's an endorsement. No buts about it.

As for being worth a read, it absolutely is not. It's a sensationalist techno-jargon navel-gazing non-story, designed purely to raise a non-issue and snare readers to bolster the BBC's click through rates. This honestly makes me wonder if BBC paid for the plug.


...sez you.


The problem here too, is that they're trying to create the uncanny valley, rather than create a non-human human.

I suspect that we fall into the valley when we commit to trying to make a human image but fail. And it has to do with humanistic movement. We never look at a painting or photoshopped image and think "Uncanny Valley". We think, "Bad Painting" or "Distorion". It's not until we see a character's movement and are asked to see something as human, that we can tell, somehow, that it's not. For example, the speeder race on Endor in Star Wars Return of the Jedi or The Rocketeer. I can't see how transitioning a face from a cartoon to a photo would reveal the UKV.


Indeed, that was a point I had meant to bring up but seem to have forgotten.

Being shown static images is entirely different than being shown moving video, much less being in the same room as something which is moving. A stillframe of a human in a weird position doesn't bug us that bad, but being in the presence of a "human" that doesn't physically behave like a human sets off our "weird-shit-o-meters" pretty bad.

This is why zombies and the like creep us out - people don't normally twitch or shamble or messily devour the dead. Likewise, a robot that "looks" human but "acts" mechanical makes our poor human brains say "Does not compute!" and we get weirded out. A smile that doesn't look organic, but rather appears forced and unnatural, is not reassuring.


FINALLY. I've been saying it for years. The uncanny valley is the doofiest dumbest popscience this side of God Particles.

And it's so obviously, trivially unnecessary.


We don't need a pseudoscientific umbrella term for these things. We can just say "It looks uncanny, because it ... [select one of the above]"


Everyone loves a term that not everyone understands. Perhaps its most appealing feature.


The closer we get to defining "uncanny valley" the more people object to its definition because of how artificial it sounds. If only we had a convenient phrase to encapsulate those objections.


Chill out, dude. The BBC did not pay for a damn plug. If you don't like the piece, that's cool. But please don't accuse me of secretly taking payola.


I didn't accuse you of anything, to be perfectly frank, but I'll at least apologize for being unnecessarily snarky.

To be honest, I think my personal distaste for BB's advertisements disguised as "posts" influenced me negatively, but that's a separate matter entirely and no excuse for insulting insinuations, so my fault there.

I suppose I just hold BB to rather high journalistic standards, and I get disappointed by what I perceive to be failures to meet those standards. Not your problem though, entirely my own beef and I need to not let it influence me negatively.


shitty motion capture, i.e. beowulf, polar express. I don't see the need for terribly much more evidence to be brought as evidence of the phenomenon.


Although why use so many descriptors, when just one rather poetically wraps them all up. One could say that the word pseudoscientific itself is semantically unnecessary and haughty when words like fraudulent, incorrect, improperly researched and poorly documented are really what we are trying to say, but as your comment illustrates, it's pleasantly punchy.