Friday, July 09, 2010

The Uncanny Valley 4: Magic, Miracles, and the Necessary Third

As many of you know, I was a tad bit obsessed with a certain theory in robotics known as "the uncanny valley" several months back. I even delivered a philosophy paper this past Spring using the uncanny valley as one way of explaining our aversion to racial passing. (You can read my series of posts on the uncanny valley here.) But, as often happens with passing interests, my
affection for the topic eventually waned-- much to the delight of my friends, I think, who had been listening to me prattle on about it for too long. But just last week, I was hosting a couple of houseguests, and one morning over coffee I had a conversation with one of them that re-sparked my uncanny valley curiosity. (I'll let that friend remain nameless, for fear of his suffering reprisal from those who've heard their fill on this whole issue.) My guest described a project he was working on that dealt with mirror neurons and whether or not they were really the cause for the affective phenomenon we call "empathy." At any rate, in the course of describing our (and other animals') apparent ability to "experience" or "feel" things empathetically-- that is, without experiencing those things first-hand-- my interlocutor described the empathizer's experience as a kind of "simulation." That is, he speculated that when we feel (empathetic) pain upon witnessing anther's injury, for example, what is happening is that our brain is neurologically "mirroring" the experience of the other, in effect manufacturing a simulation of that experience that convinces us, physically and emotionally, that we are experiencing it as well.

Now, what I found interesting about this, and what led me back down the rabbit-hole of the uncanny valley, was that it again raised the fundamental question: is it possible to have the experience of a "perfect" simulation? The theory of the uncanny valley, of course, suggests that it is not possible. As least when it comes to simulations of the human form, the uncanny valley phenomenon seems to show that our brains are structured in such a way as to generate extreme aversion to, and ultimately to reject, simulations that too closely approximate the "real." In my earlier posts on the subject, I elaborated upon this further, speculating that this is because we are cognitively "hard-wired" to protect the distinction between reality and simulation, much in the same way that we need the distinction between "true" and "false," or between a proposition and its negation. Implicit in my earlier musings on this topic, but not quite explicated, was the suggestion that, althought it may be (technologically) possible to manufacture something like a perfect simulation (in the sci-fi, "virtual reality" sense), it would be impossible to experience a perfect simulation qua a "perfect simulation." As I said earlier, if a simulation is perfect, then my experience of it will be indistinguishable from my experience of the "real" thing that it is simulating. Hence, any first-person phenomenological account I give of that experience will NOT be an account of a "simulation" but of (what I take to be, albeit mistakenly) an experience of the "real." (This is why, I speculated, if one were really being decieved by a perfect robot-simulation of a human, the phenomenon of the uncanny valley would disappear. That phenomenon requires that one be at least minimally cognizant of one's experience as the experience of a simulation. The resulting aversion to the too-close approximation of reality, I argued, was an aversion to being decieved.) So, on my account anyway, the really interesting question is not whether reality can be perfectly simulated or not, but whether or not we could ever have the experience of a perfect simulation qua simulation. So, here's a claim that I want to add to my earlier musings on the uncanny valley:

Any possibility for an account of the "perfect simulation" requires, necessarily, a third.

The person experiencing a perfect simulation, if it is indeed "perfect," does not experience it as a "simulation," and consequently will have no conceptual or linguistic tools at his or her disposal for giving an account of it as a "perfect simulation." (This is, of course, the age-old Cartesian worry, played out ad nauseum in brain-in-a-vat thought experiments and The Matrix movies.) I'm not sure that the recent neuroscientific experiments on mirror neurons or the psycho-physiology of empathy really disrupts this claim, given that even those experiments recognize that the first-person experience of pain or joy is categorically different from the empathetic experience of the same. In order for a simulation to be a simulation, it must be recognizably distinct from that which it is simulating. (Otherwise, it would just be "real," right?) So, if there is to be any account of a "perfect simulation"-- again, qua simulation-- it requires that there be an observer, independent from the person experiencing the simulation, who can still experience, understand, and articulate the difference between the simulation and reality. (Someone like the "Morpheus" character in The Matrix... or someone like the observing scientist in the laboratory full of brains-in-vats.) But, of course, if the observer-- who still realizes the "perfect" simulation as a simulation-- is the only one who can give an account of it, then... alas, we still do not really have an account of the "experience" of the perfect simulation. We have an account of the experience of "the experience of the perfect simulation." Why? Because, again, we need the ability to distinguish between a thing and its opposite, the basic law of noncontradiction, in order to give a sensible account of anything we experience. And the person experiencing the perfect simulation no longer has that ability. What that person has is, in "reality," the coincidental experience of A (a real thing) and not-A (a simulation of a real thing, that is, an unreal thing)... only that is a thoroughly irrational experience, both unthinkable and unsayable.

[Why, you may be asking, does this require a "third"? Why not simply a "second"? Well, I formulate it that way because I'm trying to stay within the parameters of the uncanny valley theory, which is not about any-old simulated experience, but the simulation of an experience between two human beings, or a human being and a very-human-like robot. So, in that scenario, I think we need three: the primary subject, the "other" whom he or she is experiencing (which may or may not be a simulated human being), and the "third" observer.]

One last thing, which may or may not make this clearer. (And which, not unrelatedly, hearkens back to a post I did a long time ago entitled Anatomy of an Illusion, inspired by the excellent film The Prestige.) I think the difference between the experience of a perfect simulation, on the one hand, and the experience of the experience of a perfect simulation, on the other hand, highlights something structurally similar to the difference between the experience of "magic" (with an understanding here that "magic" is really illusion and trickery) and the experience of a "miracle." A perfect magic trick aims to entirely veil the part of it that is a "trick," to produce the illusion of not being an illusion, and thus to make its appearance in our experience appear as a miracle. (Material objects don't just disappear and reappear with an 'abracadabra' and a wave of a wand! You can't saw a woman in half and then put her back together! Natural laws cannot be suspended! It must be a miracle!) If we ever really had the experience of a perfect magic trick, one that perfectly masked its illusion, we would, in effect, have the experience of a miracle. But, in fact, when we watch a magician do his trick, even if he is very, very good at it, we are experiencing it as an illusion, as a trick, as a simulation of something miraculous. That is why we can give an account of our experience of it as magic, and not miracle. Compare that to the experience of watching a young child in his or her first encounter with a really good magician. The child experiences the trick as real; in fact, the child does not experience it as a "trick." We adults, looking on and knowing that there is no such thing as "real" magic, can observe the total assimilation of reality and illusion that the child is experiencing, but neither we nor the child are in fact having the experience of a "perfect illusion."

It's either an illusion, in which case it is not perfect, or it's perfect, in which case it is no longer an illusion. In the parlance of The Matrix, you either take the blue pill or the red pill. The consequent implications of that choice in your experience are mutually exclusive. That's why there's no uncanny valley in the perfect simulation and, further, why there may be no "experience" of the perfect simulation at all.


Anonymous said...

I for one am on of those that would like to hear more about your thoughts on the Uncanny Valley.

But, your question, of whether I can 'have the experience of a perfect simulation qua simulation' seems to add an unnecessary intentional state in the scenario. My question is, why add "qua simulation" ? One can (in theory) have the experience of a perfect simulation, which, yes, would not be experienced 'as of' a simulation, but that doesn't mean it's an experience of the 'real', but an experience 'as of' the real, which would be definitive of the phenomenology of a perfect simulation. It sounds like your're saying that, because the phenomenology of a perfect simulation is identical to that of a real other, there is no experience of a perfect simulation. I'd say, since they're ontologically distinct, an experience of a perfect simulation is just that: an experience of a perfect simulation, and to demand that it be qua simulation seems irrelevant to the phenomenology in question. Perhaps you're saying that we can't (phenomenologically) individuate the experience of the perfect simulation from an experience of a real other, but isn't that similar to a point about one's inability to discriminate between a (perfect) hallucination of a palm tree and perception of a palm tree?


DOCTOR J said...

@Ivan (or dan, I'm not sure): I suppose the quick answer would be that I'm interested in how one might give an *account* of the perfect simulation. The person experiencing it can't give an account of it (as a simulation), and the person who can give an account of it (as simulation) isn't experiencing it. I think it the "qua simulation" clause *does* matter, because that is what seems to unveil the peculiar impossibility of giving an account of (or, perhaps, having) the "experience of a perfect simulation."

John said...

Doesn't the type of simulation you refer to have the qualities of what Wallace Stevens called the "supreme fiction", that is, a fiction that is so compelling and total, without apparent delimitation or "frame", that it is indistinguishable from the real? I think you are engaging with the question of what constitutes the basis, support, or ground of the real without asking a question that would be groundless, such as in asking "What is reality?"

If a third is required for the possibility of a (real) account of a perfect simulation, isn't this a return to or reworking of the question of possible objectivity within the framework of intersubjectivity? That is, isn't it an attempt to identify once again the "real thing" that would be the pillar of support for the framework, so as to avoid what Borges would call "contamination by unreality" that is at the heart of identification with fiction, however compelling or magical? (This theme recurs in his writings, "Ariosto and the Arabs" is one example.) If the danger of the "unreal" exists what is to stop it from implying, necessitating an infinite regress of (more or less real) supports or principles?

Finally I am interested in what you have to say about the difference between magic that would be a "trick" and an alteration of the real that would be total, a "miracle". I wonder whether and how such moments are written into the philosophical account. Hegel had his theological writings, no? And also his "spurious infinity"...