Sonic Qualia & "Scientific" Testing
But there's something about Mary. A fellow scientist, Kenneth, has secretly engineered her life so that it is utterly normal, except for one thing: Mary has never seen the color red. Everything she's ever laid eyes on has been some other color (footnote 1).
Here's the question: If Kenneth reveals his trick to her, and tells Mary that there exists a color that she's never seen, could Mary know what the color red actually looks like without looking at a ripe tomato or a red sunset? Would all her knowledge provide her with this understanding?
Answer No.1: Of course Mary could learn what red looks like. She'd ask, "Kenneth, what's the frequency?" and add that to her knowledge base. Then she'd know everything science can say about the perception of red—including, of course, what it looks like. If she still didn't know, her knowledge wouldn't be complete (but we've supposed that it is). She'd be like a mathematician who understands everything about algebra and yet finds the equation "5+7=12" foreign and mysterious. Not likely.
Answer No.2 (the right answer): No, Mary will never know what red really looks like unless she looks at something red—that is, unless red light falls on her retina. Even though she would understand everything going on in someone's head when they are perceiving red, that doesn't mean she would know, from a subjective point of view, what the color actually looks like. She'd still be missing the quale (plural, qualia) of red—the quality of redness as it's perceived.
Qualia—properties like redness, warmth, or the taste of chocolate—fall outside of science because they occur in our private experience. Scientific knowledge is not about private feelings or perceptions because it is objective and transferable from person to person. "f=ma" is not a feeling or emotion that Isaac Newton tried to share with us—it's a mathematical statement that anyone can understand because it's part of a shared language. So is Mary's knowledge of neurophysiology. It really doesn't matter how much she knows—her equations, diagrams, and descriptions of the brain were developed by scientists studying it from the outside, and say nothing about qualia experienced by a consciousness within a brain.
Mary's story throws a wrench into the mind-brain problem. The main players today are authors like Daniel Dennett, Steve Pinker, and Pat and Paul Churchland, but they might as well be Madonna and Sting. The material girl believes that mind "reduces" to the brain: All our feelings, emotions, memories, and ideas are rooted in, and caused exclusively by, the physiology of the brain. Take the brain away, and mind and its content go away, too. (Some call it the philosophy of "nothing but-tery"—not because they're watching their fat intake, as Madonna surely does, but because mind is ultimately nothing but the brain.)
But Sting says "there are spirits in the material world" and "ghosts in the machine." Take away the brain and the mind will remain, as some ghostly or spiritual substance. It's a tempting view, at least because it's so hard to imagine how a mass of neurons can become self-aware and intelligent. But you don't really clarify the issue by assigning these mysterious qualities to equally mysterious immaterial entities or spirits. When it comes to the mind-brain problem, Sting is a great singer.
Mary shows that Sting and Madonna are both right, but only partially. Even though every aspect of our mental life is fully rooted in the physiology of the brain (point for Madonna), that doesn't mean that we can understand qualia scientifically. They will always escape scientific understanding (point for Sting...well, half a point). It seems like a paradox: qualia may be identical with, and nothing but, a particular neurological state or process within the brain. But that doesn't mean a complete, scientific understanding of the brain will tell Mary what red looks like. She's actually got to see it—then she'll know it.
What does this mean for audiophiles? Take Robert Orban's letter in the October issue. He says I got things exactly backward in July's "Undercurrents," when I wrote that audiophiles (from Mars) are usually more empirical about sound than audio engineers (from Venus). He also says I'm naked, but I digress. Orban insists that audio engineers are more on the side of science and empiricism than audiophiles because they respect the results of experiments—specifically, double-blind tests. Audiophiles are "anti-experimentalists," he charges, because they reject "the results of any (blind) experiments that conflict with their worldview, regardless of the overwhelming evidence of the efficacy of blind experiments developed from experience in professional practice."
It's a compelling argument. The very success of compression schemes like MPEG AAC, Orban argues, shows that the results of blind testing should be trusted because these schemes have been developed and refined on their basis. Since these schemes "work," the methods that went into their development must work, too. But that doesn't prove that audiophiles should get on the blind-testing bandwagon. The argument actually begs all the usual questions and criticisms that audiophiles have offered against blind testing.
Suppose the subjects in a double-blind test (including me) consistently rank one compression scheme among others as sounding most similar to the original, uncompressed signal. Developers then tweak the scheme to increase its rankings even more in future tests. No doubt, the result would be a scheme that works well. The loss in sound quality will be small, the savings in bandwidth considerable.
What does this say about subjective, sighted reviewing? Nothing, because it plays no part in the process described. But it still might play an important role in the development of a product. Here we go: Over longer listening sessions, in a more familiar environment, at a different time of day, without the distractions of switch boxes or questionnaires, with less wax in my ears, when I'm not worried about my upcoming "Undercurrents" deadline, or with different components in the signal chain, I might hear compression artifacts that I didn't hear during the test. It might happen the next day or the next month, but there's no denying that it might happen. However subtle they might be, these artifacts could eventually become so vivid in my subjective listening that, had I been attuned to them during the test, my responses during the blind test might have been different.
Years ago, I never knew what "ringing" was in speakers. But a stint with some older Thiel speakers taught me. Would I have noticed these tweeter artifacts the first time I heard my speakers in a blind test? I doubt it. Those speakers "worked" very well when I bought them, and for a long time after that. But as I learned to separate the tweeter artifact from the rest of the sound, they didn't work so well for me. Listening is active, not passive. With time and experience, we learn to hear more of what a component does. (As for those Thiels, I let the pendulum swing the other way and bought some Quads.)
I stand my ground on the naked truth: The methods of audiophiles and audio engineers are complementary. We need both. Blind testing is useful for product development, but so is long-term subjective reviewing—just ask the speaker and amp designers who use it to tune their products. Your argument goes both ways, Mr. Orban.
Admittedly, audiophiles who dismiss the results of blind tests that conflict with their "worldview" are mistaken—worldviews don't have much to do with how things sound. But audiophiles don't have a monopoly on this sort of stubbornness. Those who dismiss all blind tests on principle are just as shortsighted as engineers who automatically dismiss audiophiles who claim to hear subtle, as yet unmeasurable differences between cables and components.
Nor do audiophiles have a monopoly on unproductive rhetoric. Do you really believe, Mr. Orban, that many, if not most, audiophiles are so gullible that they'll think an amplifier sounds better merely if the op-amps inside are more expensive? That may be true of some, but what is the point of so ridiculing audiophiles in general? If it's human nature for some people to pay more for a product (for whatever reason), it's also human nature for audiophiles to dismiss audio engineers who call them gullible or "anti-experimental." (That last one doesn't stick either. Experiments typically become good experiments after they've been criticized by others. That's why scientists routinely criticize each other's experiments and the conclusions drawn from them. Empiricists should be "anti-experimental" in just that sense.)
Mary sees the naked truth, too. The reason she doesn't know what red looks like, despite her perfect and complete scientific knowledge of the brain, is the reason blind testing does not have the last word on sound quality—that is, on sonic qualia. No objective language, whether of words or of equations, can effectively capture qualia because qualia are private. At best, blind testing can establish correlations among qualia on the basis of subjects' responses. ("A sounds brighter than B," "B sounds more similar to C than to A," and so on.) And it can effectively screen out biases that might otherwise cloud these judgments—it's a powerful tool, no doubt. But it can't by itself convey a direct, subjective understanding of what something sounds like.
If Mary were an audio engineer conducting blind tests, she'd know everything about her subjects—the exact state of every neuron in their brains at every moment. But, for all that, she'd never understand what the test samples sounded like unless she sat in the chair and listened. Like the rest of us, Mary would do her best as an objectivist and as a subjectivist.
Footnote 1: This thought experiment comes from philosopher Frank Jackson's "Epiphenomenal Qualia," published in 1982 in Philosophical Quarterly, 32: 127–136.