This blog is not only about me. To record my mind, I also have to, where possible, record down the thoughts of my friends who are a part of my life, mind and consciousness. And they have made this post possible. So, I thank them for allowing me to post their views here.

My best friend responded to Hofstadter’s article Sounds Like Bach (mentioned in my earlier post). As Chris observed, this may just be the longest email my best friend, the tortured artist pursuing THE DREAM (As Chris puts it. I would say “doing THE DREAM”), has written to his wife, Chris and myself at the same! The record-breaking email, ladies and gentlemen:

Boon,

You finally sent an article of real interest to me….. lol (just teasing)

Actually i did find it very interesting. I read the article first and then listened to some mp3’s. The EMI pieces pose interesting questions, but i think Hofstadter’s fears are somewhat unfounded. Here are my reasons:

1) The EMI software, like other AI software that try to emulate (perhaps surpass) a human ability (such as all those chess programs) does not function purely on their own. A Human hand is always required to set what parameters the software is to work within. Whats more,
should the software not work, should it produce a result that is incoherent, it is again a human hand that ‘corrects’ this through re-programming or tweaking of the software. Thus, I suggest that the pieces generated by EMI are not so much a reflection of Artificial
Intelligence, but rather the musical intelligence of the programmer.

2) The pieces themselves are very clever, and do, as claimed, play in the style of Bach, Chopin, Beethoven, Mahler etc…. However, in a way they do this almost too well, and even not knowing the true source, one gets the impression that whoever created these compositions took
extra pains to create a sense of ‘authenticity’. My point is, that they resemble too much the works they try to emulate. I am certain that Beethoven could not have written the sonata styled after him because it is too much like the Moonlight Sonata, and sounds if anything, like a cheap imitation. Put it this way, had I not known of this software, I would have guessed the composer to be a human emulating Beethoven. It not so difficult for a human to have done what
EMI did. It is far more difficult to create something of meaning without using pre-existing models, which is what Beethoven did. A computer cannot create something from nothing.

3) Hofstadter fears that music may one day be reduced to a syntactic pattern, and that his idea of the mystique of music destroyed. This makes me wonder if he is thinking clearly about what music is. Of course, music is an emotional outlet and has mystical qualities to it,
and it those attributes that we prize above all in music. However, music has always had its ‘scientific’ side, from the Ancient Greeks until today. And from that side, the analytic point of view, emotional results are effected by tangible, concrete techniques. Taken one step further, we can say our emotional reactions to a turn of phrase, or the sound of horns, etc… are psychological reactions to an auditory stimulus. Take the idea of major vs. minor sounds, for instance – these are human reactions to otherwise indiferrent stimuli. The art of composition is really nothing more than the art of arranging auditory stimuli to produce ‘meaningful’ responses in a humanly coherent way. Therefore music is, in a way, through its study, already reduced to a
syntactic pattern, a series of techniques and systems designed to produce a human emotional response.

4) All that EMI does is arrange in a syntactically coherent manner, gestures and idioms that famous composers are known to have favoured. Is it any wonder then that the EMI pieces sound ‘geniune’? But they are of course, fakes. Clever ones, but fakes nonetheless. It is as if one were to collect various quotes of Shakespeare, rearrange them to fit syntactically, sew them together to produce a ‘new’ work. Or perhaps to take the eye from a Picasso portrait, join it to the mouth
from another, and so on… to create a new ‘face’.

Anyways, I could go on and on… but I think thats alot to chew on for now. Tell me what you think…

Gotta go make some arrangements now….i think i may use my computer
to help….LOL

cheers,
MCC

Chris’ response:

Heh, you know what Sifu, you only just forced me to read the article, cuz I usually receive Boon’s mail when I’m at work when I’m strapped for time and energy, and when I get home I’ve got even less energy and slightly less inclination.

I really shouldn’t let my job sap all my energy.

Hey, my colleague told me a joke yesterday: There are only 2 professions in which you can be considered an expert without having any formal professional qualifications. The first one is journalism. The second is prostitution.

Well, my take on the Hofstadter article is that: – EMI music right now is just a pale imitation of the real thing. Cut
up the good stuff and reassemble cunningly. – And, despite Hofstadter’s worries, I feel that’s all it’ll ever
likely be, no matter how cunning it gets.

Now, I feel obliged to post my thoughts on this too. But the problem is I don’t have much thoughts on EMI. But here goes anyway:

First, Hofstadter confesses, articulates and discusses a fear that has become common place since the industrial revolution (no mean feat, and this is what makes his article interesting). The fear of us not being as special as we want to be. The fear of reductionism and materialism (excuse me for using these terms careless, without definition), where the romance (illusion?) of mystery, the stuff of our inner worlds and experiences, the music that inspire and move us are somewhat reduced to a musical Turing Test. I think, to be frank, no one would be surprised if they heard the music from EMI and were told that they were direct human products. The real surprise is to be told that the music is an indirect human product, from a software.

Second, the key question is whether the fear of humans being more shallow than we thought is justified. My answer is a qualified “Yes”. I think there are serious constraints to our biology that can be manipulated easily, whether by humans or non-humans like porn websites or computer programmes, which we can expect to be far more sophiscated in the future than we can currently imagine. The notion that our aesthetic preferences are solidly grounded in biology and have a neural basis is a key idea behind neuroaesthetics (for more info, see links from mixingmemory, Financial Times and Washington Post). My first and most powerful introduction to neuroaesthetics was this passage from Ramachandran’s Reith lectures:

What about abstract art? What about Picasso. What about semi-abstract art? What about impressionism, what about Cubism? Van Gogh? Monet? Henry Moore? How can my ideas even begin to approach some of those artistic styles?

To answer this question, you need to go and look at ethology, especially the work of Niko Tinbergen at Oxford more than fifty years ago. And he was doing some very elegant experiments on seagull chicks.

As soon as the herring-gull chick hatches, it looks at its mother. The mother has a long yellow beak with a red spot on it. And the chick starts pecking at the red spot, begging for food. The mother then regurgitates half-digested food into the chick’s gaping mouth, the chick swallows the food and is happy. Then Tinbergen asked himself: “How does the chick know as soon as it’s hatched who’s mother? Why doesn’t it beg for food from a person who is passing by or a pig?”

And he found that you don’t need a mother.

You can take a dead seagull, pluck its beak away and wave the disembodied beak in front of the chick and the chick will beg just as much for food, pecking at this disembodied beak. And you say: “Well that’s kind of stupid – why does the chick confuse the scientist waving a beak for a mother seagull?”

Well the answer again is it’s not stupid at all. Actually if you think about it, the goal of vision is to do as little processing or computation as you need to do for the job on hand, in this case for recognizing mother. And through millions of years of evolution, the chick has acquired the wisdom that the only time it will see this long thing with a red spot is when there’s a mother attached to it. After all it is never going to see in nature a mutant pig with a beak or a malicious ethologist waving a beak in front of it. So it can take advantage of the statistical redundancy in nature and say: “Long yellow thing with a red spot IS mother. Let me forget about everything else and I’ll simplify the processing and save a lot of computational labour by just looking for that.”

That’s fine. But what Tinbergen found next is that you don’t need even a beak. He took a long yellow stick with three red stripes, which doesn’t look anything like a beak – and that’s important. And he waved it in front of the chicks and the chicks go berserk. They actually peck at this long thing with the three red stripes more than they would for a real beak. They prefer it to a real beak – even though it doesn’t resemble a beak. It’s as though he has stumbled on a superbeak or what I call an ultrabeak.

Why does this happen?

We don’t know exactly why, but obviously there are neural circuits in the visual pathways of the chick’s brain that are specialized for detecting beaks as soon as the chick hatches. They fire when seeing the beak. Perhaps because of the way they are wired up, they may actually respond more powerfully to the stick with the three stripes than to a real beak. Maybe the neurons’ receptive field embodies a rule such as “The more red contour the better,” and it’s more effective in driving the neuron, even though the stick doesn’t look like a beak to you and me – or maybe even to the chick. And a message from this beak-detecting neuron now goes to the emotional limbic centres in the chick’s brain giving it a big jolt and saying: “Wow, what a super beak!” and the chick is absolutely mesmerized.

Well now what’s this got to do with art, you’re wondering?

Well this brings me to my punch line of about art. What I’m suggesting is if those seagulls had an art gallery, they would hang this long stick with the three red stripes on the wall, they would worship it, pay millions of dollars for it, call it a Picasso, but not understand why – why am I mesmerized by this damn thing even though it doesn’t resemble anything? That’s what all of you are doing when you are buying contemporary art. You are behaving exactly like those gull chicks.

In other words human artists through trial and error, through intuition, through genius have discovered the figural primitives of our perceptual grammar. They are tapping into these and creating for your brain the equivalent of the long stick with the three stripes for the chick’s brain. And what you end up with is a Henry Moore or a Picasso.

Back to my second point, if you, my reader, are still with me. As far as we have biological constraints that can be easily manipulated to produce certain reactions/emotions in humans, whether by computers or other humans, I have no doubts. The question is what else is left? Is there something deeper? Is there anything left over in our constitution that cannot be manipulated or duplicated by computers? How much more? To the extent I don’t know, my “Yes” is qualified. My hope is that there is a lot more left. The future will tell whether this hope is justified.

Third, I don’t think that just because EMI cuts and combines from its input (whatever those may be), that lessens the authenticity of the music it produces. At a very fundamental level, I think that is how humans compose too, we cut and combine from a series of music notes. Could a machine achieve the same facility and “creativity”, “feeling” and “authenticity” as a human being in future? I really hope not. But let’s see.

Fourth, what I would really like to know is whether there is a software out there that can improvise to music. Not that this would add to the discussion, but because I find this interesting.