I agree with your original question ("things are CREATED to serve a purpose. What is that purpose?"), but I completely disagree with your assessment of how "The mind can be expanded upon."
The brain is NOT comparable to a digital computer. If it truly were, we would have artificial minds already. You need concrete proposals for the entire system to show that brains can be improved upon, and you don't have it. You're simply hand-waving, but philosophical musinngs about "possibility" based on ignorance (i.e. the lack of a full systematic proposal) is meaningless here.
an order of magnitude higher than most estimates of the power of a human brain
Comparing numbers of instructions that can be executed is meaningless. The number of instructions needed in order to execute a program varies based on the hardware architecture. The computers you're talking about are general purpose. Brains are not. The hardware is optimized to execute the types of operations necessary for things like object recognition, memory storage, etc.--exactly the tasks it's been "designed" for.
The architecture of the hardware dictates how 'efficiently' you can program something. For example, in your system, memory is stored outside of your 'processor'--in other words, it takes cycles just to retrieve a piece of information to process it. Another example, since your system works with a single processor, you're doing everything serially. The brain works with parallel networks. In order to simulate this parallel execution, you're going to have to swap.
You can't compare hardware architecture like this. The architectures are so different, the software is not related. Executing the software on each has very different costs. There's no reason to think that your single processor system is going to be able to functionally simulate the important features of the brain (see below), let alone how many orders of magnitude of instructions would be added in order to actually program it.
Neurons are slow.
Yes, but you're looking at individual features of a fully developed system where some bad features exist because they support the good features. You can't criticize the architecture of the brain without offering a better system. Above, you simply propose a single processor. No mention of what software is running or necessary hardware features (storage, execution, resource and power supply mechanism, maintenance system), and necessary features (plasticity, performance with degradation, generalization procedures, solving the binding problem, blah blah)
Your system will have drawbacks as well and, since we're not even close to constructing a self-sustaining system with a human-like mind, it's clear your system will be far inferior to the current human one.
...
Back to the fact that neurons are slow. Neurons are slow because of the biology behind it. In order to make a faster system, you're going to have to use different materials. How many silicon brains can you make? How do you maintain them? What happens if your single processor has a problem?
We lose neurons as we age, rather than adding on additional brainpower.
I wouldn't say this is a property of creating the human mind, but of creating humans which are mortal.
If we did add on additional brainpower, we'd LOSE information. To simply add additional neurons to existing networks simply degrades the performance of the network. You'll lose memories, skills, etc. You'll become more like a child again--easier to learn things, but you know less. It would take work to "re-learn" what you already knew before adding more "brainpower."
There's other ways to add more brainpower, but given the lifespan of the organism, it doesn't make any sense. It's unnecessary. So first ask the creator why we're mortal. We have our current set of brainpower because it's more than enough for the life span that we have.
Neurons cannot be directly "reprogrammed" by our high-level consciousness.
This doesn't have any meaning to me. Why would this be good? Are you proposing that it's better to be conscious of every neuron in your brain?
The software you're talking about doesn't work as step-by-step source code. It's meaningless to "reprogram" a single neuron. Neural networks don't store high-level information locally, but rather distributed across a network.
We can't retain our memories indefinitely.
What makes you think that this is a good thing? Retaining memories indefinitely is expensive in resources. How is it useful? It's not a hard problem to solve computationally with neural networks; just produce more neurons and don't share information across them. If this was cost-effective and useful, no matter the designer (evolution, God), it would have been done.
Propose another system where you can have this kind of storage (you'll probably propose an unlimited store? How often will you have information come in through the eyes? Since retinal neurons fire what, 40 times per second, but you want to increase it, maybe you'll get up to 1GB per second of visual information to store? Of course you'll have to write a program where the hardware does no computations on the visual input, otherwise you won't have a correct record of events. That means a whole lot more cycles for memory retrieval, as every time you retrieve memory, you'll have to reprocess it from scratch in order to make any sense from it. Blah blah...)
We can't transmit thoughts from one brain to another.
That's the purpose of language. In order to transfer information directly between brains, you'd have to guarantee that the information is stored in the same data format. In order to guarantee that, you have to remove the flexibility of coding information based on your past experiences. In other words, you have to hard-code your data format instead of allowing the organism to use self-updating hardware to both code and restore information. Unless you were going to pass both the data AND the decryption device along... and what data do you pass across?
I don't see a system better than language. Maybe you can propose one?
...
All right, I'm done. I'd be happy to listen to your response to this semi-rant. Just taking some pent-up frustration about brains vs. computers out on you. I don't feel that I'm doing a good job, so you should have plenty of loose ends to pick apart here.
Ben
™ Version 4.2