Ain’t No Mountain High Enough, Ain’t No Uncanny Valley Low

February 12, 2017 § 5 Comments

Image result for uncanny valleyImage result for uncanny valleyImage result for uncanny valley zombiesImage result for uncanny valley robot

Looking at the images above, how do they make you feel? Disturbed? Uncomfortable? Maybe even scared? The four pictures shown are all humanoid with a face,mouth, eyes, and every feature we as humans have. However, each photograph makes the viewer feel that something is off. They look human, yet there is something about them that is not entirely organic. The second photo shows the most realistic looking human out of the four, right? Well, head on over to this site and see if you feel the same: Her head moves with the direction of your mouse, and she also changes facial expressions. Yet, though she looks like any other human being in a still picture, her facial movements are far from being perfectly human.Thus leading us down to the uncanny valley.

Image result for uncanny valley chart

The term “uncanny valley” comes from roboticist Masahiro Mori. In 1970, Masahiro Mori hypothesized that there comes a certain point when an object that is human like becomes strange due to its lack of familiarity. In his translated words, “[T]he appearance is quite human like, but the familiarity is negative” (Pg. 2). This leads humans to feel unease at the the unfamiliarity produced by these creations. Thus, Masahiro Mori had a message to designers out there: “I recommend designers take the first peak as the goal in building robots rather than the second. Although the second peak is higher, there is a far greater risk of falling into the uncanny valley” (Pg. 3). He implored designers to take this hypothesis into consideration and that we should not be too ambitious as to design a robot that is as human as possible, for even missing that mark by a small percentage will lead to the discomfort of the uncanny valley.

Scientists have been wondering why this phenomenon happens. One reason may have to do with perception of experiences. Can you trust your eyes? Take a look at this video and see just how many of these mind tricks you can out-see:

Now, this may seem unrelated to perception. However, there is a crucial point in the video. “Our brains and eyes have evolved to see, but our vision makes assumptions based on learning, memory, and expectation…” (2:10-2:14). We as humans, through our everyday experiences, have come to expect what a real human looks, acts, and feels like. We have a set criteria that makes humans human because of the information we received from what we see day to day. So when something comes across our eyes that doesn’t meet our expectations of what we think it’s supposed to look like, it leads us to the uncanny valley.

Ayse Pinar Saygin, Thierry Chaminade, Hiroshi Ishiguro, Jon Driver, and Chris Frith conducted a study on people’s Action Perception System (APS), which includes the temporal, parietal, and frontal areas of the brain, and looked at fMRI (functional magnetic resonance imaging) repetition suppression. Repetition suppression is a phenomenon where there is “reduced neural response to a repeated stimulus” (Pg. 415). Therefore, “[P]ositive suppression means there [is] less response to repeated stimuli (Pg. 416).” They hypothesized that ” the uncanny valley may, at least partially, be caused by the violation of the brain’s predictions…” (Pg. 414). We assume a certain object will move a certain way because that is what our life experiences have taught us. However, when our APS sees that the movement is not what we expected, a type of “error” occurs (Pg. 415).

The study had 20 participants reacting to 3 different agents (though one participant’s data was excluded from the end results): a robot with robot-like movements (A), an android with robot-like movements (B), and a human with human-like movements (C). Both A and B used the same agent, but they were dressed differently. A was made to look like a robot in which the mechanics of the face and body were revealed. On the other hand, B was dressed with human-like qualities, such as skin, hair, and clothes. This robot/android was called Repliee Q2 and was modeled after the human in agent C (Pg. 415). The participants would then be monitored with fmri to see how their brains would react to videos of these 3 agents moving in transitive (picking up, grapping, wiping, etc.) and intransitive (waving, bowing, nodding, etc.) motions (pg. 416).

The results showed that APS reaction to A and C were quite similar. However, there was a clear difference between these two and B. The participants’ brain scans for agent B showed much more repetition suppression. Though this experiment was not designed to explain the uncanny valley, it leads us to believe that there may be a connection to our brain’s APS and the phenomenon created by the uncanny valley. Because the actions of the Android, Agent B, who looks like a human, did not present the actions participants thought it would enact, a prediction error occurred in their APS (Pg. 420).

This shows there is a sort of discomfort when introduced to things that are familiar yet do not perform as expected, a discomfort hypothesized by Masahiro Mori. The objects, motions, and scenery we see everyday are the norm for us. But what does the uncanny valley implicate for our daily lives? Can you really trust your eyes? Perhaps there will come a time when we really can make androids that look as real as we do in which we cannot see any difference and finally surpass the uncanny valley. But is this a good thing or not? What does it say about humans and humanity? Is it something that can be created and imitated, even if it is not organic?


Ayse Pinar Saygin, Thierry Chaminade, Hiroshi Ishiguro, Jon Driver, and Chris Frith

Uncanny Valley by Masahiro Mori



Tagged: , , , , ,

§ 5 Responses to Ain’t No Mountain High Enough, Ain’t No Uncanny Valley Low

  • resnichm says:

    I want to relate these questions of the uncanny valley to other industries besides A.I., to innovation. When I read this case study and thought about the separation in our brains between the robot, the human-like robot and the human, it reminded me of the invention of the airplane. When the Wright Brothers were attempting to create a machine that flies, they first looked to birds and made bird-like airplanes with wings that flapped. This, of course, didn’t work on a larger scale, so they scrapped the idea of mimicking a bird and instead made something with fixed wings, a propulsion system and moveable controls. They tried to make the plane equivalent of B, a bird-like plane, but opted for A, an innovative machine.
    I’m not sure what good would come of achieving B, a perfectly humanoid robot, but history shows us that A has suited industrialization well in the past. I look forward to our class discussion, I think others could shed some light on the value of B.


  • elizabeth1315 says:

    When thinking about the value of a human-like robot, I think it’s important to include in the discussion that we are still at the beginning stages of the possibilities for robots. Therefore, engineers shouldn’t limit their designs to include exclusively A-like models. The Wright Brothers failed with a bird-like plane, sure, but that didn’t stop them from eventually getting it right. In fact, they likely learned lessons from their many original attempts that they eventually applied to their successful attempt. Additionally, it makes sense to start with imitation, whether that be a bird-like plane or a human-like robot, when you are creating something new. After all, it’s what we know. Even the robots that don’t look like humans have human-like features, or more broadly, features of living creatures. Machines that do manual labor have extended “arms” that move in the same way our own arms and legs move with joints. It’s incredibly difficult to create something (or even imagine something) entirely new without first imitating something else.

    Aside from this, robots that look like humans may be able to perform certain functions that would be unique to them. Non-human robots are great for performing tasks better than us or tasks that we can’t complete, but human-like robots would be great for human-specific tasks. Human-like robots have already been shown to potentially help autistic children. It is certainly likely, then, that there may be many practical and positive applications for human-like robots.


  • gsbendik says:

    When reading this blog post, the first thing I thought about was the movie Coraline. It’s a children’s movie that is actually quite disturbing, even for adults. In this animated film, a young girl named Coraline finds a small, hidden door in her new home that she climbs through to meet her “Other Mother,” which is an almost exact replica of her mother only she has buttons for eyes. This image is so unsettling to viewers that we immediately begin to infer that the Other Mother is secretly evil even though she comes across as amazing the first few times Coraline sees her. This idea of the buttons for eyes is also in keeping with the understanding of eyes in The Sandman reading we had to read for our upcoming class. The old saying that, “Eyes are the window to the soul,” that is embodied in The Sandman is also embodied in the movie Coraline and since the Other Mother has no soul, she also has no eyes.
    These almost human humanoids are more disturbing to humans than completely nonhumans because their inorganic state is much more apparent. It feels unnatural and wrong in many contexts. I do think it is interesting that autistic children have responded so positively to humanoid robots and I do think these positive interactions complicate the issue and make it impossible to say an adverse reaction to humanoid robots is more than likely to occur. However, I do think that intention is also very important. When creating a humanoid robot that’s goal is to help autistic children, clearly, the inventor has a relatively altruistic or at least good-hearted intent. Often, when we see humanoid robots gone wrong in films and books, the creator has impure intentions, such as to emulate God or to exercise their hubris. I think this is important to note because it demonstrates how often the motif of a humanoid robot is being used not to say that they are innately bad, but to make a comment about human nature overall.


  • Emmie Kline says:

    I’ve always found the idea of humanoid robots highly disturbing. Personally, even as a child, I never liked art installations, such as wax figures at the Madame Tussauds museums, or other creations that resemble humans too closely. On the other hand, I’ve never had an issue with robots or creatures that are obviously non-human. As mentioned in other comments, I think this natural aversion to humanoid robots is, in part, due to the characteristics that are so obviously inhuman, such as imperfect facial movements, but I also think that the fact that these humanoid contraptions remind us that we are close to creating a perfectly human robot. Sometime in the future, scientists will be able to create a robot that seems perfectly human, right down to its reactions, facial expressions, and movements, like something straight out of a science fiction story or film. At that point, will we know who is human and who is a robot? It reminds me of the clones in Blade Runner in certain ways, as even Deckard begins to refer to the replicants as “people” and even falls in love with one.
    Personally, I don’t have a problem with most robots; most people my age do not know life without them. But I do believe there is a fine line between innovation and possible trouble in trying to create a perfectly human robot. In my opinion, there’s no need, as we can create robots to serve a specific function or perform a certain duty without giving them a human likeness. Maybe Frankenstein is getting to me, but I think trying to replicate human life is always a slippery slope…yet one humanity has an endless fascination for.


  • Olivia Peel says:

    I had a conversation with one of my friends about this phenomenon back in December. We had just seen the new Star Wars movie “Rogue One”, in which they digitally reanimate the late Peter Cushing through CGI. I thought that this was creepy just on principle, but my friend was very unsettled by the CGI Grand Moff Tarkin. I remember her saying something about how “they can never get the eyes quite right when they try to do these things”. She then sent me this link to the Uncanny Valley trope on tvtropes:
    tvtropes is a great website for geeks like me. (I would warn anyone reading this that this website has the tendency to suck people in for actually hours, so go forward with caution) In a nutshell, it describes common tropes and then connects them back to shows/movies/novels/etc. Frankenstein, Rogue One, The Polar Express (referencing your image above), and Doctor Who (with the gangers I talked about in class last week) are all listed under this trope. What I find interesting is that, while Rogue One and The Polar Express are listed because they were somewhat unsuccessful in trying to recreate humans, Doctor Who and Frankenstein use the trope to their advantage. Designers/writers can use the uncanny valley to communicate that we *shouldn’t* quite trust something, or that something isn’t quite human.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

What’s this?

You are currently reading Ain’t No Mountain High Enough, Ain’t No Uncanny Valley Low at Science/Fiction.


%d bloggers like this: