If beauty doesn't require God, why should morality? (Bite me Dr. Craig.)

[bennyboy]

It's not, whats all that biochemistry about then? Ever encountered a floating "mind"? Ever really dove into the architecture of the brain?

A little bit in uni while getting a psych degree. But I don't think you're getting the point. Mind is not a category of anything else. A sphincter is a specialized muscle made of specialized proteins, etc.-- it is derived from other things, and is therefore a specific member of a more general category.

I suppose you could say that the human mind is a complex of more fundamental qualia. In that case, you'd want to identify the most basic possible "atomic" qualia, which aren't constructed of any others. But how would you know them for what they were even if you COULD make them?

I find it fascinating that ultimately, what we are looking at with regards to dendrites and axons (under our current model) are chemical and electrical NAND gates, which we understand to be incredibly powerful from our experience in comp sci. Universal gates, you can build any function out of them-it's just an issue of how many you've got (and we have alot).

If that's literally true, and if qualia are a category of brain function, then we should at least in theory have the ability to create an entity which actually experiences qualia, rather than just seeming to.

But how could we know that was the case?

Now, of course...maybe it's just coincidence that they're set up that way, maybe they don't do anything of the sort - but they do one hell of an impersonation of machine logic- and what we can measure about them forms a fairly simple machine language. It would almost be a waste if they weren't NAND gates. Of course, a more efficient machine could be built with specifically chosen gates- but biology (and especially evolutionary biology) doesn't have a selection of gates to choose from - so we would expect to find redundant "brute force" architecture in any "evolved circuits". The one thing you'd need, with a generalized series of NAND gates that aren't predefined and redundant to the point of being built individually for every single task - would be a set of program NAND, that "chose" which serious of gates to use based upon the task at hand(it takes a different number/arrangement of NAND to emulate other gates). In this context it's easy to see one of the many possible functions of "self awareness" or "mind" and also that it isn't necessarily "other" . This is the reason that we're searching for ai, the ability to monitor ones own system, however flawed or inefficient, yields computational returns - and it doesn't have to be anything more that sufficiently arranged NANDs itself, as we've built self referential machine systems out of NAND already, too many to count.

I've done some programming with ANNs. I'd say two things about that-- 1) given a complex enough networked system with a strong feedback mechanism, and I'd say the evolution of useful patterns would be the norm, rather than a big surprise. It's slightly frightening to think what a decent AI programmer with decent hardware could teach a computer to "think" by taking internet activity as sense data. 2) none of this leads me to think any particular physical system could actually experience qualia, or that we could ever had a non-arbitrary test to determine whether it did or not.

People are actually parsing the architecture of the brain at a basic level - that of possible individual logic gates.

Careful, there. Given how neurotransmitters work, this is an artificial (i.e. statistical) oversimplification-- unless you have a link that indicates otherwise?

The answer to that is simple - anything can be used to perform a logical function- the only requirement is that the function is adequetely mapped. It doesn't matter whether you use water, stones, or little woolen blocks covered with a mysterious substance known as "redstone" in a children's fantasy game (virtual logic machines built inside of a program that resides in a machine, wewt). Machine logic is just a way of manipulating the observed behavior of "things" to perform a specific task.

Let me ask you this. Let's say you encoded an .mp3 song on a beach using black and white stones as bits. But the problem is, you are the last person on the beach, and you are so lonely that the act of encoding is your last hurrah-- once done, you commit suicide.

Assuming that no information about the encoding process remains, are those stones still an .mp3 song?

Simply put, this explains qualia as a service that the machine provides, self referential, and as a modifier for computational power (without such a system machines can't compare apples to oranges, or offer multiple outputs and meta-analysis of those outputs).

Conflating self-referential data processing (or any other kind of data processing) with qualia is essentially begging the question-- you are defining qualia in a special way, and your model subsequently seems to make sense. But this conflation is unsatisying philosophically. Given that simple mechanism can explain all the functions of the brain, there's no reason why that mechanism would actually experience in the way that I experience redness or emotions. We can (at least hypothetically) make machines to replicate human function, and with ANNs we can make them self-referential. But there's nothing about supposing such a system to actually experience qualia that will improve our observations or understanding of the system.