I don't get the sense the author really understands the existing literature.
In fact, I think the author has misunderstood the main point from Mori's 1970 paper[123], the counter-intuitive relationship between "affinity"[z] and human-likeness -- the uncanny valley itself.
For a specific example, consider the discussion of Mori's "famous graph" near the beginning of the article.
> the graph assumes you have an objective and linear way to measure similarity to humans. In reality, we rely on human intuition to know what is more or less similar, and this leads to wildly inconsistent determinations.
Mori's graph isn't a data plot. It's a drawing to illustrate the concept (the "valley") in an intuitive way.
I am certain that Mori understood very well (just from reading the paper, haven't a clue about anything else about him) that human perception is not an objective or linear function.
The article circles back to repeat its own talking points many times, is full of strange and unsupported claims, and the author keeps circling back to their (mis)understanding that "more human" appearance should result in a proportional "less unease", i.e. the "myth" Mori was busting way back in 1970.
This prompted me to read the paper again, so thanks for that!
Agree, and there's also further research in babies that have demonstrated the "facial recognition" or "more human appearance = safer" isn't all that true, as babies will look at congruent objects (two blocks over one smaller block over a round background) just as much as faces (Macchi Cassia et al., 2008).
I agree. TFA calls out the “fundamental weakness” of the theory as a lack of specificity in definition and measurement, then goes on to make further opinionated statements as far as a proclaimed definition without referencing the much deeper history of the uncanny in literature (e.g. The Sandman, Frankenstein) and philosophy (Jentsch and Freud).
Yeah, the constant repetition was actually lulling me into a deep boredom. And everything discussed was anecdotal. No actual research to support anything being said. Who wrote this? Or upvoted it?
The author had it right up until they tried to generalize "uncanny" to equate to "loss of control". It's not the same thing. "uncanny" happens as a result of your brain not being able to dismiss a potential threat (and thus you remain on alert and can't relax fully). That's not the same thing as a loss of control.
The author also fails to go into our evolved genetic psychological makeup (which varies according to what "stock" you hail from): Certain kinds of differences are more threatening than others because survivors over time have tended to naturally find them more threatening - you see this often with the insectoid-looking holes in sliced lotus root for example.
Threat. I find artificial voices I engage with over the phone at work flip my threat response, because the emotional affectation is patently false. I feel the same way about certain humans who follow a strict script sprinkled with false concern for my feelings. It’s like junk food. You can’t get anything from the interaction that helps you to communicate.
Ok, I see your point. Yes, it's a specific flavor of uncertainty.
The ones you listed are about outcome uncertainty (anticipation). And uncanniness is about the nature of things.
I think it really has to do with a having a noticeable line between an articulated face and simpler body for human like characters.
Thanos's face was really well animated. Even though it was mocap, he face felt like it was part of a living thing. When he talked, the rest of his body moved with it. I think the oversized head help give more room to transition. Contrast that to the faces from Cats. There is a clear delineation between the human face pasted onto the CGI cat body.
Clu's face also looks pasted on to a body. All the de-aging tech suffers from this because they are usually pasting a CGI face over a real body, and the micromovements don't match. Whatever tech smooths out wrinkles also seems to reduce fidelity.
Animatronics suffer from the same issue. Often the faces are highly articulated, but the rest of the body is just a frame covered in plastic.
Worf and Gamora were both real actors in makeup, so they still had all the human micromovements. And anything that is fully animated doesn't have the same delineation.
I wonder if this is why people are uncomfortable with autistic people. Some subtle details of behavior are missing and they can't identify what because all the superficial features may be present. They'll tolerate it for a little while but eventually get sick of not being able to relax and accept the person as fitting into any desirable or known type of personality.
Perhaps moving to a foreign country makes it easier because you're obviously different - like a cartoon character or upside-down Thatcher.
I'm high enough functioning, but still get certain minor anxiety pangs around those who are more actively presenting. I feel a bit bad about it, especially when I know from personal experience that they're mostly normal and well-meaning enough.
I can confirm it does. Although I moved to a country with more direct auti friendly communication (USA to NL) and only recently realized I'm likely on the spectrum and this heavily motivated my desire to move here where communication is much more direct.
The people who would do the learning aren't the people suffering from being misunderstood so they won't.
I feel like I've improved at detecting fake smiles though, and am more comfortable seeing resting bitch face or real grumpy face than a fake smile that might be hiding hostility.
Things that look like snakes like those weird mimic moths or a coiled garden hose in tall grass can also trigger that uncanny feeling to me. I suspect serpent recognition is coded into many animals at a very deep level, and anything that trips that recognition will seem uncanny.
If you have cats you can trigger this response with many of them, using a leather belt.
If you hold the belt at the buckle end and then lay it on the floor while rotating your wrist back and forth, the belt will writhe and curl in the manner of a snake.
Many/most cats will instantly respond to this and they will try to strike the belt while simultaneously keeping as much distance as they can -- but still compelled to keep trying to strike the threat.
Affective Blindsight detection of snakes is real and supports that at least the primates have that; there's plenty other animals that freak out at snake like objects (cats, elephants, etc)
The example of "Clu" in TRON: Legacy is a bad example. In the movie they are inside a simulation and everybody besides the real humans do look a bit off, on purpose. So 3D generated young Jeff Bridges fits perfectly. It does add to the movie that they look "not completely real".
Great soundtrack by Daft Punk by the way. The story is linear but it's still an impressive movie.
Uncanniness is mostly about misrecognition, bound to familar concepts. Something that is familiar to us will get processed lazily by our brains. We trust it in a certain sense. Things get uncanny when it is just at the tipping point of being a new thing that we understand as such and a familiar thing that we know. And because it is at a tipping point this can chaotically oscillate back and forth between familiar and unfamiliar within a short timespan. The resulting overlap of emotions feels "spookier" than a unfamiliar thing alone would feel.
The author starts by giving 6 tiny thumbnails 3 of which are fine and 3 of which are uncanny. I honestly don't see how the 3 marked as uncanny are uncanny. They also don't mention from which films/media those images are.
I recognized Worf, Gamora and Thanos but not the others and with the tiny thumbnails I don't see what's uncanny about those 3 pictures.
For what's worth - we have 2 perception systems, top down (essentially pattern recognition) and bottom up (requires sense-making). When we experience something that doesn't fit the first - automatic - our homeostasis is diminished as we experience less orientation/control ("what's this?"). Surprise is the emotion that activates this reassessment of our approach to the given stimulus. Quickly followed by our attention and then cognitive appraisal of the given stimulus that wasn't classifiable by the top-down automatic system.
The "uncanny" experience is given by this dissonance - "it looks like a face while it's not a face" - and it is just a shape of surprise as emotion. Surprise is a short lived emotion, quickly followed by a different one. The unease associated with the idea of "uncanny" is when we experience disgust as the second emotion. Not everybody experiences that. Some of us experience another emotions, like anticipation/curiosity.
The need for orientation, or to understand what is going on, is tightly interwoven with a sense of control. To have clarity about a situation is to have a sense of control and options become more apparent. When we don’t know what is going on in our environment, there can be high levels of stress and anxiety.
For a specific example, consider the discussion of Mori's "famous graph" near the beginning of the article.
> the graph assumes you have an objective and linear way to measure similarity to humans. In reality, we rely on human intuition to know what is more or less similar, and this leads to wildly inconsistent determinations.
Mori's graph isn't a data plot. It's a drawing to illustrate the concept (the "valley") in an intuitive way. I am certain that Mori understood very well (just from reading the paper, haven't a clue about anything else about him) that human perception is not an objective or linear function.
The article circles back to repeat its own talking points many times, is full of strange and unsupported claims, and the author keeps circling back to their (mis)understanding that "more human" appearance should result in a proportional "less unease", i.e. the "myth" Mori was busting way back in 1970.
This prompted me to read the paper again, so thanks for that!
[123] https://ieeexplore.ieee.org/document/6213238
[z] https://japanese.stackexchange.com/questions/11496/what-is-a...