Stephen Pinker (again) and the Decline of Art

[PianoIsTheRemedy]

1. Couldn't it be the case that SOME behaviors are "innate", some are caused by experience, and some are a mixture of both? Why is it one or the other?

2. I see glaring evidence against Pinker's broad polemic. One particular point that I feel compelling relates to orphans who are devoid of human contact and nurturing at birth. They develop profound learning disabilities and social problems. Indeed there may be innate "machinery" FOR behavior, but it requires nurturing to develop.

[2bsirius]

Very interesting take on Pinker. I had never thought of his analysis of art that way. I can't find the book now, and I don't remember the subtlety of his argument well enough...BUT what you say sounds reasonable. For example, in cultures where "classical forms" in architecture [i.e., the Greeks and to a lesser extent the Romans] wouldn't those influences themselves, as you pointing out, mediate toward a certain art aesthetic independent of raw genetics? It makes sense to me.

[randyhelzerman]

I couldn't have understood conferencereport's characterization of Pinker's take on art correctly---art is declining because we aren't writing poems that rhyme, and rhyming poems are built into our DNA? That couldn't be it. Surely Pinker's read "Leaves of grass?"

[randyhelzerman]

lets do some math. the human Genome contains roughly 3 billion base pairs. 2 bits per base pair, plus or minus junk dna, dna used for things like muscle contraction--how much information is left over to control our thoughts, prejudices, behaviors, etc?  1 Gigabyte would be a huge overestimate. Now, how much information can the human brain store? A three year old child makes about 1 million neural connections every second of her waking life. The adult brain has 100 billion neurons, (cont)

[randyhelzerman]

(cont, to conferencereport) and untold trillions of connections between them. So sure, I'd say that we have some inborn behaviors, etc, but for any given behavior, there is at least a 4 order of magnitude greater probability that it was a learned behavior than an inborn behavior. The similarity of behaviors across cultures just goes to show that despite surface differences, human cultures are far more alike than they are diverse.

[2bsirius]

Nice to run into you here, Randy. It makes me feel a bit like the cognizanti. Btw I think I probably spelled cognizanti wrong. It's hard to be the cognizanti if you can't spell it! ;0)

[randyhelzerman]

LOL 2b I can't spell worth a damn :-)

[CousinoMacul]

Randy,

While I agree that most of our behaviors are learned (but withing constraints), your calculations are meaningless. I could do a similar calculation to *prove* that a program that draws a Mandelbrot Set doesn't contain enough information to describe the Set, therefore the M Set's complexity must be *learned*.

You are forgetting that it is epigenetic mechanisms that build brains and they can do a lot with little. Also, what about animals with complex unlearned behaviors?

[randyhelzerman]

Hi CousinoMacul, but it is DNA which makes those epigenetic mechanisms too. Sure, it is hard to calculate exactly what the storage capacity of the human brain is (do we count 100 billion neurons? do we count the connections between the neurons? ) but there really is no ambiguity about how much information is stored in the human genome: its 2 bits per base pair, max, and that's nowheres near enough information to specify even the simplest human language, let alone a whole culture.

[CousinoMacul]

As to DNA making epigenetic mechanisms, yes and mostly no. The point of my fractal example (I could have used Turing patterns or cellular automata) was that you can generate complexity from little information. You seem to think that there should be some sort of bijection between the information content in the genome and that in the phenome. That is just not the case.

[CousinoMacul]

As for the no ambiguity in the information stored in the genome, what about overlap? We know that happens in the real genome. For example, the following 5 bit "DNA" sequence 12345 can generate the 15 following "RNA" sequences 1, 2, 3, 4, 5, 12, 23, 34, 45, 123, 234, 345, 1234, 2345, 12345. And that's not counting non-sequential "RNA" sequences (which does happen in actual transcription).

[2bsirius]

I hate being treated as if I'm not in the room when the real intellectuals are talking. What am I: rancid cheese? I know lots about Mandelbrot Sets...OK...Can only do some of the easy iterative math...Epigenetics mechanism? You're beginning to sound like me, CousinoMacul. Great comment even if it wasn't addressed to me. The complex unlearned behavours is intriguing.

[CousinoMacul]

I acknowledge that K is in the room and I let her know that I've responded to Randy again. ;-)

[randyhelzerman]

Hi CousinoMacul, I don't feel the force of the mandlebrot example. The mandlebrot set contains no more information than the smallest program which generates it (you know about Kolmogorov complexity, so this point should be elementary to you). Sure, overlap/epigenetic info shows a very efficient use of the information storage available, but still, the calculation is unambiguous: the maximum amount of information storable in 3 billion base pairs is 6 billion bits. (cont)

[randyhelzerman]

(cont, to CousinoMacul) This is a remarkably small amount of information: I've uploaded only a few hours of videos to YouTube, for example, but surely I've uploaded more than 6 billion bits (less than a gigabyte!!) and that has also been in very efficiently compressed mp4 files.

[CousinoMacul]

Of course I'm familiar with KC, but if you know nothing about the program which generated the M set, and charged with finding the amount of information it contained, you would probably drastically overestimate it. Which is the exact mistake that I think that you're making now by estimating the information coded in human behavior, then trying to draw a bijection between that and the information in the genome. Which brings me back to the animals with complex inherited behaviors. :-)

[randyhelzerman]

Hi CousinoMacul, so I gave an example: if you even videotape 6 hours of my life (my behaviors) and compress it with the best compression methods we know about, you've already used up more information than what is in our genome. What's more to argue for here? We couldn't specify the grammar of English in less than a gigabyte of information, let alone a lexicon of any size--how is it not blindingly obvious that human culture is not genetically transmitted? (cont)

[randyhelzerman]

(cont, to CousinoMacul) w.r.t. complex animal behaviors, what do you think is the most complex nonhuman animal behavior, and how would you go about counting how much information it would take to specify it?

[CousinoMacul]

OK Randy, now you're just straw manning me and getting away from my original criticism. I never argued that all human behavior (or even that most) is genetically determined. In fact I even said as much (check my original comment). I just criticized your information bijection example.

As to your question, I don't have a method.

[randyhelzerman]

I always argue in good faith Javier. I may very well have misunderstood your point, but I've never strawmanned anybody in my life.

[CousinoMacul]

Let me just remind you of the arguments I had with Jeff (az) a while back, but my definition of straw man does not require intentionality but simply arguing against a position that your opponent does not hold. I didn't think that you were intentionally misrepresenting my position. :-)

[randyhelzerman]

Well thanks for giving me the benefit of the doubt by assuming that I was too dumb to understand your position then :-)

[CousinoMacul]

What you call dumb I call stuck in a frame. ;-P

[randyhelzerman]

Hi 2b, you're not rancid cheese, and I'm not a real intellectual, I just play one on youtube :-)