« New Audiocasts | Main | SimPolitics »

The Second Uncanny Valley

second uncanny valley.jpg

The "Uncanny Valley" is the evocative name for the commonplace reaction to realistic-but-not-quite-right simulated humans, robotic or animated. Most of us, when encountering such a simulacrum, have an instinctive "it's creepy" response, one that is enhanced when the sim is moving. Invented by roboticist Masahiro Mori, the Uncanny Valley concept is typically applied to beings (broadly conceived) as they become increasingly similar to humans in appearance and action.

But what about beings as they become less similar to humans -- following the path of transhumans and, eventually, posthumans?

An article in the latest issue of New Scientist (subscription required) prompted this question. Thierry Chaminade and Ayse Saygin of University College London began to investigate how the Uncanny Valley phenomenon worked, and performed brain scans on people encountering simulacra of varying degrees of human likeness. They found spikes of activity in the parietal cortex.

This area of the brain is known to contain "mirror neurons", which are active when someone imagines performing an action they are observing. While watching all three videos, people imagine picking up the cup themselves. Chaminade says the extra mirror neuron activity when viewing the lifelike robot might be due to the way it moves, which jars with its appearance. This "breach of expectation" could trigger extra brain activity and produce the uncanny feelings.

The response may stem from an ability to identify - and avoid - people suffering from an infectious disease. Very lifelike robots seem almost human but, like people with a visible disease, aspects of their appearance jar.

Clearly, such a reaction does not require that the observed "human" actually be sick, only that its behavior and/or physiological characteristics seem a bit off. This could, conceivably, include human beings with "enhanced" characteristics -- "H+" in the current jargon.

Science fiction visions of space-adapted posthumans with hands for feet or wings for low-gravity flight would obviously seem at least "a bit off," but the enhancements need not be that radical. In fact, it's possible -- even likely -- that the less-radical changes would end up being more disturbing. Enhancements to optical capabilities might change the appearance of the eye. Improved neuromuscular systems might make everyday actions -- grabbing a coffee cup, picking up a child, even walking along the street -- look unnatural. Accelerated cognition might make verbal interactions disjointed, even bizarre.

As long as these changes fall into the broad ranges of current human variety, we'd be unlikely to see an unusually negative response. But if they are clearly outside the realm of the "expanded normal," and if they have external manifestations that are readily identifiable, it may very well be that the reactions of unmodified people -- and perhaps even the reactions of other "H+" individuals! -- are significantly more negative than one might expect. In this scenario, the enhanced person wouldn't just seem weird, he or she would seem wrong.

If this is possible, then it has profound social and political implications for transhumanist and other H+ advocate agendas for human enhancement technologies.

For example, if the typical reaction of unmodified people to enhanced humans is "that guy really creeps me out," it may be easy for opponents of these technologies to generate a legal and cultural backlash.

Similarly, if the gut reaction to a moderately modified human is to see him or her as no longer human, political struggles could get very ugly very quickly.

It's unlikely that the first generations of human enhancement technologies -- which would most likely just be adaptations of therapeutic medical technologies -- would engender this kind of response. But if we follow the logic of the human enhancement model, we will at some point over this century start to introduce changes to the human physiological and behavior model that will fall well outside the realm of human variability. It's possible that we'll have enough other kinds of simulacra and non-human persons in our midst that we'll take such modifications in stride, and have no qualms about keeping the transhumans in the human family.

But it's also possible -- arguably, more possible -- that the emergence of significant modifications to humanity will trigger deep responses in the human brain, ones that we may very well not like.


I find the assumption here that there's a "linear" dimension to "human likeness" a bit problematic, don't you?

The implicit assumption is that robots in the uncanny valley are "less" than human, while there's "another" valley on the "more than human" side. It seems to be an attempt to prescribe a single direction for "posthumans" to evolve in, when the reality is more likely to be a diaspora of morphological diversity. It sounds like an attempt to tack arbitrary moral values onto morphologies, perhaps in an attempt to justify some kinds of "creepy" over others.

There's only one uncanny valley; things or people are either like what you're used to, or not. "transhumans" or "posthumans" would be progressing back to the the left, away from 100% "likeness".

I wanted to be clear on a distinction between things made/evolved to be like humans, and things made/evolved from humans. You're right, of course, that the function of the uncanny valley is the same either way. Further, I agree that the H+ era is likely to see a diversity of morphologies, both physical and cognitive.

There is an implicit progression in the traditional uncanny valley concept, however, since the disturbing aspects of simulacra get more profound as we are able to develop more sophisticated versions. Layering the various forms of H+ onto the existing map would run counter to that implicit progression. I would imagine that if we put the various proto-human hominid species onto the uncanny valley map, they would (likely) align more or less in evolutionary steps. Having enhanced/engineered new human variants follow the opposite developmental direction would be confusing.

By no means is it an attempt to justify or morally privilege some kinds of "creepy" over others. You should know me better than that, Nato.

I think of it as advice, jump don't inch away from human form, its less confusing for those of us working with only a few pounds of monkey brains.

I think Jim nails it there.

What if there are unknown peaks over the "healthy person" area and there are posthumans who are unbearably sexy and charismatic? Angelic, even.

*They just kinda show up while you're doing your taxes.... "Rise and come with me." "Yes Lord!"

Angelic, sexy, and charismatic, and as likely to mate with a current-day human as we would be willing to shtup a Bonobo.

Olaf Stapledon describes post-post-post- post-post-post-you get the idea post-humans living on a terraformed neptune 2 billion years in the future as looking like lions, tigers, and bears, which would certainly excite the furry contingent.

I do know you better than that, Jamais; the researchers in the linked (and mostly paywalled, apparently) article I know less about - to say nothing of those who would interpret the piece.

Is that your graph, or theirs?

It's my graph, based on the one in the Wikipedia entry for Uncanny Valley. The New Scientist article doesn't mention trans/posthumanism and UV at all.

Reminds me of how the concept of the monster is used socially to normalize disruption caused by the percieved difference of outsider groups.

There's an essay by Eileen Joy where she describes female Chechen suicide bombers-"Black Widows," as both intimate and alien, and provocative of a similar response to the one described above.


Wonder if the neurological phenomena you describe stems from an evolutionary response to cultural and ethnic difference?

To Jamais: I'm new to transhumanism and whatnot, so pardon any stupid questions. Is the Borg (Star Trek) trans/post human? Where does he/it/they (not a Trekkie, either) fall in this Bell curve with a twist?

Your graph looks very scientific, but what exactly are the x and y coordinate to represent? By what do you meter empathy? Or human resemblance? I think this is related to Nato's post.

Also, why are apes sometimes 'creepy' and sometimes our brothers (fall both high and low on your graph)?

Finally, I think I would caution people from comparing the distinctions that will arise (and have arisen) from trans/post-humanist advances to those based purely on ideology. I suppose that is to say, every xenophobia is distinct and will be resolved under its own conditions. The challenges e.g., cyborgs will face in society will be radically different from Jews in early 20th century Europe or from Africans in America.

I think Hegel would be of use here. Knowledge always comes right before the end of an era, always looking back. Perhaps we here, sitting on the dawn on the post-human era, look back upon yesterday (and even today, we are always looking back, even at ourselves) with the knowledge of the "Uncanny Valley." But this era will end when it absorbs its opposite--its fear of the non/near-human (whose roots are fed by the soil of Plato and the water of Christ)--and overcomes both that and the "Uncanny Vally" which represents the Truth of that fear.

To Reid: Examining the historical and geographic diversity of cultural and ethnic identities ("Whites" or "Caucasians" were until recently as variegated as European and American communities, from families to nations. Today, "blacks," as are recognized in the West, are, in many communities, acknowledged to be ethnically and culturally diverse. Whether an individual belongs to this culture or that is relative to the time and place of the judge, not the subject of the judgment.) and attitudes (Cross-cultural and ethnic relations range from physical violence on an isolated or institutional level, social inequities, and "double standards" to erotic mystery and promise--even indifference.) reveals that if Us/Them, the recognition of another as Other, as strange or foreign, as alien, has an evolved neurological basis, then this basis must/should be as heterogeneous as the phenomena it produces (sorry about that sentence). I highly doubt, then, that the "perception of difference in outsider groups" best describes evolution-based neurological (as opposed to other) phenomena. Instead, I would argue, this describes socio-historical phenomena. What do you believe could be gained by your proposed line of investigation?

To Bruce S.: Excellent point. I think the only think we can say about the future is that we know nothing about the future.

Sorry about the long post, I'll be back for responses--please respond!

I think "uncanny valley" is explicable as cognitive dissonance to category breaking. Almost-humans aren't a special case. People would also find an almost-cat or an almost-house uncanny. The motif of almost-X for some X is a staple of Sci-Fi art.

The hill after the valley is when differentiation is strong enough to create a new category. Therefore: when becoming transhuman, name what you have become.

I think you're almost right. But we don't need to name or conceptualize the Pixar-lamp in order to anthropomorphize it. In fact, this "lamp" breaks both the categories of "lamp" and of "human." So why respond to it with empathy and not cognitive dissonance? The same is true of many anthropomorphic cartoon characters, which are neither human nor animal (see "Over the Hedge", "The Lion King", "Lady and the Tramp," etc.--all of which star creatures that are almost-human and almost-dog/raccoon/lion/whatever, both in physical appearance and in behavior).

How do you these cases in relation to your comment?

I agree that there must be a Second Valley of the Uncanny, or more specifically a more elaborate postulation of the valley. I don't see a need to make a new valley altogether, just to broaden the field on the first on. Who made this graph of The Second Uncanny Vally?

I think about the idea in this post often.


Creative Commons License
This weblog is licensed under a Creative Commons License.
Powered By MovableType 4.37