Tuesday, October 20, 2009

The Uncanny Valley 3: φύσις and τέχνη

As I recounted in my first post on the uncanny valley, I learned of this phenomenon from a student in one of my classes. I can't remember the exact context of his bringing it up-- it had something to do with our knowledge of the outside world as contested, and then proven, by Descartes in his Meditations-- but I do remember my student claiming something like that we could never really be totally deceived by a "virtual reality" simulation, because our brains are "hard-wired" to know the difference between reality and its simulation. The uncanny valley "proves" this, he claimed.

Now, I'm almost always suspect of claims about the so-called "hard-wiring" of our minds, but I am especially so when it comes to what seem to me like socially-constructed categories. However, I will admit, of course, that there are some rules and structures that appear to be inviolable in human thinking. The law of noncontradiction, for example. Or the fact that all "experience" is known in space and time. Or that God speaks to humanity through Johnny Cash. (That last one, especially.)

Now, my claim in the original uncanny valley post was that there must be some kind of meaningful and significant dividing line that intersects the bottom of the valley. In the context of the uncanniness of robots and the like, I think that intersecting line distinguishes between the human and the non-human. But I think that line could, and does, represent other major categorical divisions in other contexts, as I suggested in the second installment might be true when we think about race or racial passing. I think we could chart a similar phenomenon with gender and sexuality, too. The more I think about it, the more the "uncanny valley" seems to represent a kind of basic structural phenomenon that is not only iterable but, dare I say it?, hardwired in human thinking. If this is true, then that means that the uncanny valley can't really be about "the human" (or any race, gender, or sex category of "the human"), but rather must be about some other superstructural distinction of which those other distinctions are substructurally derivative.

As I tried to suggest, though incompletely and somewhat unconvincingly in the original post, I think the important distinction being made in any uncanny valley is between the real and the apparent. Or, in different terms, the distinction being made is between that which is "made to appear" and that which "naturally appears." In Greek philosophy, it's the difference between physis (φύσις, that which has a beginning in itself) and techne (τέχνη, that which has a beginning in another). So, the root of the aversion that we feel to the coincidence of the familiar and the unfamiliar is, as I originally claimed, an aversion to deception, to taking the simulation for the real, to conflating the "like" and the "is." Products of techne can be "made" to be very much LIKE physei onta, but we have a deep and abiding (perhaps "hardwired") commitment to maintaining the distinction between the truth and its (even VERY CLOSE) approximation.

Over on his blog, Anotherpanacea has a different take on the uncanny valley (which I highly recommend reading). He focuses on the singularity of the human face and he claims, among other things, that our aversion to things like bad plastic surgery is motivated by the fact that plastic surgery introduces ambiguity into our ready-to-hand taxonomies of facial recognition, thus producing (through techne, I would add) the impression of the familiar in the unfamiliar, and vice versa. He contests my claim that our aversion is at its root an aversion to deception, because he claims that the experience of the uncanny is prejudicative or prereflective.

I'm sticking to my claim, though, because I think in any experience of the uncanny there is at least some kind of minimal reflection or judgment happening. I think I can justify this with the following example: if we were actually deceived by a simulation, that is, if we really took it to BE what it was only LIKE, then we wouldn't have the experience of the unheimliche. We would just have the experience of the "real" (familiar) thing. So, somewhere in the hidden recesses of our consciousness, it seems to me, the experience of the uncanny MUST involve a simultaneous and coincidental experience of two categorically different things (the familiar and the unfamiliar, the real and the simulation, the truth and the approximation). Our aversion to this kind of ambiguity is tanatmount to an aversion to deception. Now, I think AnPan is right that it doesn't quite rise to the level of complex reflection and judgment where one might make the possibile-deception thetic to oneself. However, if we want to explain how something like the uncanny valley can be applied to several different categorical distinctions (which I obviously think it can), then we need some manner of articulating the more fundamental violation of the difference between "being" and "being-like."

So, although I hardly think that things like racial or sexual categories are "hard-wired" into our brains, and I would be willing to allow for the possibility that the human/non-human distinction isn't either, I think the interesting thing about the uncanny valley is that it shows how even in the case of socially-constructed categories our commitment to the reality/simulation difference produces the same kind of cognitive dissonance when it is breached. If we take "whiteness" or "female" or "human" to mean something REAL, whether that "reality" is "true" or not, we find ourselves attracted to the simulation of that reality only up to a certain point, after which we are averse to the danger of mistaking the false and the true, the manufactured and the natural, the "like" and the "is"...

... whatever "it" is or is like.


Dr. Trott said...

I'm sure if you talked to Dr. Grady about this then this has already been raised, but, I'll bring it up anyway. You know the example that Kant uses in the 3rd Critique of the fake bird? So we like the sound of the bird song, highly repetitive though it is, until we discover that it was a mechanical bird and a recorded song and then we no longer think it was beautiful. Kant's point I think is about the beauty in nature that is compromised when we see it is from techne. So that might be one way to consider the uncanny (I almost called it the unhappy! ha!) valley.

anotherpanacea said...

"The more I think about it, the more the "uncanny valley" seems to represent a kind of basic structural phenomenon that is not only iterable but, dare I say it?, hardwired in human thinking."

I think the metaphor of hardwiring runs aground here, and we might want to look at language acquisition issues and Chomskyian universal grammar. The closest approximation might be a computer's 'firmware' or maybe just its operating system: the stuff that tells the computer how to interpret the software, that mediates between the neural and the experiential. That's why the work on implcit bias and facial recognition is so important to me: if racial bias is something that can be 'trained away' then we can tell an institutional/pedagogical story of responsibility for racism and white privilege, whereas if it's just 'hardwired'... well, then, we're blameless, aren't we? More to the point, this would suggest a corrective strategy in contact, desegregation, and affirmative training.

At the same time, one of the disputes I think we're having is whether this is purely a 'social construction' or something like an institutional construction or the negative consequence of one or several evolutionary biological trait. We have a lot of good fMRI data on what's happening when a brain sees a face: I'm not sure we should jettison that in favor of a more phenomenological story.

Take a look at the Karl MacDorman lecture series I linked: it's only about twenty minutes long. His efforts to measure uncanniness in various settings shed a lot of light on the phenomenon and the sorts of things that are likely to trigger it, and while I think there's an analogous problem in race or human/animal distinctions, I don't think the data can be explained using even a minimal 'reflection' on real/appearance, but rather we need something like 'expected/unexpected.'

Another reason the line probably shouldn't be set at the human/inhuman divide because there are already androids and dolls we can identify on the near (familiar) side of valley. We're successfully deceived by some robots, now, which on your terms would suggest that we feel less-deceived by a well-crafted robot than we do by a person with bad plastic surgery. Our reflection on that ought to suppress the comfort we feel at the robot, but it doesn't.

One last issue: I think it's a real question in moral philosophy what role the disgust reaction plays in various moral judgments, and I'm unsatisfied with the various psychoanalytics accounts. This is a place where experimental philosophy seems to be doing better work in identifying reactions that are grounded in disgust and telling plausible stories as to why disgust might have an evolutionary role to play in survival. Let me know what you think on this issue... I'm trying to write a paper on the role of disgust (and to a lesser extent in-group preferences) in political intuitions, so I'd appreciate some general feedback. (For instance, I don't find Kristeva helpful.)

DOCTOR J said...

@Dr. Trott: It was my conversations with Grady that really forced me into seeing this as a problem of how we think about "images." He didn't specifically mention the birdsong example from Kant's 3rd Critique, but I've certainly been reminded of that myself, in much the same way that you articulated.

@AnPan: There's a lot in your comment, so I just want to remark on one particular thing (for now). I don't think that a case in which we actually mistake a simulation for the real thing that it's simulating is a case of the unmheimliche. If we REALLY take a copy for the real thing, then we don't experience it as "familiar/unfamiliar," we experience it as the real thing. The uncanny always involves a kind of coincidence of recognition and non-recognition (I like your categories of "expected/unexpected"), which don't coincide in an actual "error."

anotherpanacea said...

If we REALLY take a copy for the real thing, then we don't experience it as "familiar/unfamiliar," we experience it as the real thing.

That's a good correction... I was being a bit sloppy. But if we can simultaneously *know* that something is non-human and feel homey and familiar around it, then what does this mean for passing and white privilege?

emma b. said...

What about the distinction between imitation as mimickry and imitation as becoming?

I'm thinking about language learning here, or "faking it till you make it" with a skill... at what point do I shift from playing at someone who teaches philosophy to being someone who teaches philosophy (and in the meantime does the uncanny valley apply to my students... am I the object of their revulsion?)

That old phusis techne distinction... in terms of human and becoming and development it seems to break down a bit. Is going to the gym a techne? What about if I built these amazing muscles climbing trees?

The Clapp said...

The evolutionary history of this trait seems of value for more insight. If the uncanny valley continues to show up repeatedly, it seems reasonable to suggest that we have some tendency, either hardwired or cultured. This article seems to support the former: http://www.wired.com/wiredscience/2009/10/uncanny-monkey/

First seemed to me that this would be a good evolutionary trait for distinguishing a friend from foe: the evil twin from the saint sister. Perhaps the reason technology, and robots in particular, freak us out, is because of the film industry's portrayal of our demise at the hands of such creatures. This is obviously conjecture, but there seems no doubt that this is linked to some sort of survival instinct: we fear that which appears to be similar but is actually dissimilar, because those sorts of distinctions are important for deciding which berries/mushrooms/organs to eat.

This also seems connected to our particular fear of certain movies. I do not know about you, but the most real scenarios are always the freakiest (and seem to pose the most threat to my body and mind). The intentional amateur home video makes movies like the blair witch project, cloverfield, and paranormal activiy more like a documentary of real horror, which is truly terrifying. I know that it is not real, but am confronted with real emotional responses, which provides the uncanny valley. The question that leaves me with is:

Is my horror directed at the perceived reality, or at my recognition of its untruth and unease with that tension?

And what the hell leads people to enjoy that tension and/or horror?

Does this have anything to do with our fascination with the completely abnormal and unknown?

Dr. Trott said...

I was put in mind of the uncanny valley today as I was reading "Robots That Care" an article by Jerome Groopman in last week's New Yorker. He explains that old people with Alzheimer's were more likely to do tasks to keep them sharp with robots than with computers for the very reason that robots seemed more like them. The computer science professor running the study said, "It's counterintuitive, because you would think that if you give very old people robots they might be put off by how alien they are. But it actually seems to be less alien than the computer." The uncanny valley suggests the what is counterintuitive is precisely the opposite -- the old people should be put off because the robots aren't(!) alien, but they are not. Probably because the robots are alien enough.

anotherpanacea said...

A couple more articles for you, Dr. J. Apparently monkeys share our distaste for the almost-but-not-quite similar Other.


Lovely to see y'all this weekend!

The Clapp said...


How do you create links in comments? I was trying to link the monkey article in my post, but ended up with plain text.....


DOCTOR J said...

@Walter: In order to put links in the comments, you have to use some basic html code. Instructions are here.

Dr. Trott said...

I didn't read to the end of the New Yorker article before I commented last -- they go on to mention the uncanny valley!

Lorenzo said...

I enjoyed anotherpanacea's post, thank you for recommending it.

A lot of apparent prejudice can be explained as communication issues: that we do not know quite how to interpret/what expectations to have, etc about people who are different/unfamiliar. Which would fit with the link in AP's post about certain forms of recognition training reducing prejudice.

So Michael Jackson's plastic surgery is both deceptive in your sense and throws out some of our use of face singularity in AP's sense.