Example Drift: Uncanny Valley

A good way to think of whether something falls into the Uncanny Valley: Does it requires context?

Example of Uncanny Valley: Bob's lips look rubbery, which makes his smiling creepy.

Example of not Uncanny Valley: Bob is really creepy when he smiles while killing people.

One requires outside context; the second person does not fall into the Uncanny Valley, since the only cause of creepiness is the juxtaposition of the situation and the person. The first does not require the situation around it; he could be smiling while strangling someone or smiling while watering flowers, but the smile itself looks wrong.

edited 26th Mar '11 11:28:20 PM by Scardoll

If something is happening in the very same scene, it's not context.

If it would be, even Bob's rubbery lips would be context, because they only work in the context of being on his face, smiling.

Outside of the wiki, the term is about "almost human things are scarier than non-human things or fully humans." That's it.

Robotics and animatronics are just two common applications of it, but not the complete definition of the term.

From the first two google pages of results for the term:

http://www.slate.com/id/2102086/

http://www.arclight.net/~pdb/nonfiction/uncanny-valley.html

http://seedmagazine.com/content/article/uncanny_valley/P2/

These are just from the scientific articles trying to define Uncanny Valley, don't even get me started on how random forum-dweller nerds tend to use the term.

It looks like your interpretation of the core definition is just one of the many.

edited 27th Mar '11 2:29:30 AM by EternalSeptember

Well, before we can discuss the split, we really need to work out a definition.

I think we all want to to two things here, which is to preserve the original definition of Uncanny Valley as much as possible, and to give a name and place for examples of a more generalized version of Uncanny Valley.

If we take into context that the original article was written by a roboticist for roboticists, then perhaps we should take a look at something Sakku said on page 5: [quote]Maybe the problem here is that the term is being used to describe a similar idea but from different perspectives. To a roboticist Uncanny Valley is about making robot appear not quite human, in both looks and demeanor. To a graphic designer Uncanny Valley is about photorealism and surrealism, and images that aren't quite human. To a writer Uncanny Valley is about demeanor because that's the primary way he expresses the "not quite human" aspect in his characters. And so on. Could this be a term that is used to mean a more specific subset of a broader trope depending on your perspective and background? [/quote]

I think we should use Uncanny Valley for the original visual meaning in an effort to preserve it, using criteria something to the effect of "if you saw this character walking down the street, would you think it was creepy?" That way we use a mundane setting and context to eliminate cases of tropes like Dissonant Serenity, since the latter involves extreme situations and context.

The other trope can contain examples of the more generalized version. It can easily accept in universe examples and non-visual descriptions. It might even be a supertrope or sister trope to the more specific creepy-personality tropes we already have.

The key here, I think, is expectation. In the more general sense, we expect a thing that seems human in most cases to be a human in all cases. So we can see how the roboticist talking about robots to other roboticist applied that concept and we can also see how the general population, having caught onto the general concept underlying the original definition but lacking a term for it, used the original term to describe their own examples.

If all of these perspectives equally exist, there is no reason to give preference to the one that happens to be first.

As we commonly bring it up in such topics, Tsundere was coined to mean Defrosting Ice Queen, and the writer who coined Manic Pixie Dream Girl considered it a type of Shallow Love Interest.

This doesn't force us to keep the original Trope Namer interpretation, if the new interpretations are becoming more common (as it happened with Tsundere), or if they are more practically tropeable (as it happened with MPDG). Uncanny Valley has both of these reasons to keep the current description.

The more I think about it, the more convinced I am that even a split is unneccessary. It has 20,118 inbound links, it's one of the most popular pages we have, and the wicks are all used consistently with the description, they would only be incorrect if you would want to limit it to the old interpretation that Mori Masahiro used for it.

True, but that's the impression that I got from the people here.

If people want to keep the original definition then split it along those lines.

If not, then any other split is even more arbitrary.

Edit: The way I see it is that we basically have to decide first whether we want to keep any of the examples that don't fit in with the original description at all. We seem to, at least for the sake of compromise. Bearing that in mind, we logically have three choices: we can either rewrite the description and lump it all it, or keep the original definition and make a split for the stuff that doesn't fall in the original description, or draw some other distinction and make a split that way.

Personally, I think the third option is much weaker than the first two. Now, the major difference between the first two is how much we want to preserve the original definition. If we're all okay with having one page with both the original definition and how it's used and applied in general, then we can take the first option. If not, and we don't seem to be agreed on this point, then we split off what we want to keep into a different page with its own name and a link to Uncanny Valley. Names and such can be decided afterwards. The specifics of the split-off trope can also be decided later, since we can clean up the main page in the meantime.

Of course, the third option is viable if we can agree where to split it. But I don't see how any other split is better than along the original definition.

I wish I had a flow chart for this. Basically: Do we want to keep the examples that don't fit the original description? Yes. Do we want to lump them in one page and rewrite the description? We can't agree on this point, so no, tentatively. Then we should split it. How do we split it? Along the original definition? This is the best I can come up with. Along some other line? This seems like a weak option.

If we can't find a place to split, then we need to lump it and rewrite the description.

edited 27th Mar '11 6:58:49 AM by arcbeest

The problem with that quote is, that the writer bit was never in the original definition of Uncanny Valley. It's one of the things that's decayed into it at this site. It's not actually used for the trope anywhere else. Just the robotics and CG definitions.

To me, the proposed split looks like just splitting the In-Universe version away as a separate trope. We could do that, but I mean, it's not like we have to. We can just mark Uncanny Valley as an Audience Reaction soft-fplit In-Universe examples on its page, and allow In-Universe examples on main pages like all the other Audience Reactions.

I'm not trying to split In-Universe from Audience Reaction, so much as the trope from people misusing the trope.

So, "Just the robotics and CG definitions" on the web count as correct use, because even though there are other interpretations all over the web, they are something else than "Just the robotics and CG definition", so those don't count?

edited 27th Mar '11 8:58:15 AM by EternalSeptember

Because CG and robotics are where the term was invented and defined. It's how they're used in text books and formal uses. People misuse things all over the internet. It doesn't make the right.

Yes, it does, if those "people on the internet" include roboticists who claim that there are multiple Uncanny Valleys based on our psychological reactions, comic book theorists who claim that a realistic drawing can fall into it, and furry fanfic writers who say that anime... err...

Anyways, the point is, that if so many people from so many fields are using a more generic interpretation of the term, and they dominate the Internet hive mind, what gives the legitimacy of Mori Masahiro's version? That he was the first?

As the examples of Manic Pixie Dream Girl and Tsundere show, being the first to vocalize a concept, doesn't give you exclusive right over it's future development.

Not to mention, that even Mori Masahiro wasn't that clear on what falls into the valley as you claim. He was the one who included zombies and bunraku puppets in the original chart.

And he didn't include those creatures for their personalities but for their appearance and movement which fits with my definition.

People say all sorts of things on the internet and it's very hard to tell the veracity of most of it. I will stand by the formal text book definitions of it that don't include personality.

edited 27th Mar '11 9:20:20 AM by shimaspawn

Of course, personality is not an example of the trope by any mainstream definition.

That's what I was talking about with the writing thing vs. the robotics/cg thing. I was referencing something given in a post two posts above mine. I think you took my words out of context of the post above them. I was saying that no mainstream definition ever based it on personality.

edited 27th Mar '11 9:42:41 AM by shimaspawn

You were talking about splitting "people misusing the trope" from the actual trope.

The trope is not misused for personalities, it is only used on this site for things that are too realistic to be humans, but not unrealistic enough to be objects.

The trope is frequently misused for personalities invoking a similar vibe thanks to tropes like Uncanny Valley Girl making it seem like that was the focus of the trope as well as an inability of tropers to understand the idea of action vs. personality.

Frequently? Did we ever do a wick check?

The op had a short one, but it was mostly about not photorealistic things falling into the valley, (that the mainstream interpretation actually allows), and "Very slight behavioral examples" like frozen expressions, that are closer to actions than personalities.

I don't see how you could possibly provoke the Uncanny Valley Effect in an audience through writing. You could write an in-universe example, certainly, where the characters themselves are creeped out by the inhuman movements of a being that looks human. But the medium of writing simply cannot itself get that instinctual reaction that is at the heart of the trope.

This is restricted to mediums/works that approach photorealism. Anything else is not going to provoke the audience reaction that we're talking about here. It has to be done through the way the "person" looks and moves. A "person" in the Uncanny Valley is always in the valley, no matter what the situation. It's not dependent on circumstances or plot.

I'm not really sure why this is hard to grasp.

I've never seen a mainstream definition that lists the non-photorealistic stuff as anything other than "stuff that doesn't fall in the valley." It's frequently on the approach ramp going up to the valley, but it's never in the dip of the valley itself.

All the tree of my earlier quotes included that, two by clearly stating it, and a third, the roboticist, by claiming that there are multiple uncanny valleys, depending on psychological causes.

The Uncanny Valley is subjective: A video game character that is more detailed than the other chararacters around him, can be called an example of the Uncanny Valley for being too detailed, but in a game where everyone looks like him, and one character is as undetailed as the normal people of the earlier game, the latter would be an example, for being slightly off from the game's default.

Your first quote insists that it's photorealism that causes the issue. You even highlighted this fact.

Your second article when the quotes aren't taken out of context shows only photorealism as falling into the uncanny valley. It stresses that it only happens when they try too hard to duplicate human and miss a little detail. It seems from context that anime touches upon the issue in narrative, but not in art style.

Your third article also stresses photorealism as the cause when the full article is read and bits aren't just quoted out of context.

Huh, it seems like all your sources agree on photorealism as a necessary component. There are multiple sources of the effect, but they all come from divergence from a photorealistic base.

edited 27th Mar '11 10:55:40 AM by shimaspawn

Of course, photorelism is a necessary component, it's at the end of the chart. A fully photorealistic human is already out of the Uncanny Valley, and less realistic characters than that are in it.

But how far do they have to from it? Mori Masahiro's interpretation doesn't draw an objective line, only vague statements that it's "very close to" or "99% close to" photorealistic.

Out of the tree articles, two state that drawings can still be in it, and a third speculates that there can be multiple Uncanny Valleys depending on the context.

Your own statement that drawn characters can't be in it, isn't supported by anything but your own arbitarily made-up interpretation.

The Uncanny Valley is a failure at photorealism. It's getting too close to the concept but not hitting it. All of your articles agree on that point. Yes, one says that drawings fit in, if they attempt photorealism and fail in a small way.

And yes, one says there can be multiple small valleys, but it still posits the reason for them as a failure at photorealism. Don't just use lines out of context.

edited 27th Mar '11 11:24:50 AM by shimaspawn