|
|
Post by general313 on Nov 15, 2017 16:04:47 GMT
I tend towards the idea that hardware is irrelevant. . .All the scientific evidence so far suggests that the brain operates very much like a very sophisticated system of interconnected neural networks. I suspect that red qualia properties are determined by the structure of a neural network and not at all on the physical details of how that neural network is implemented. The nervous system has to guide your body through a world filled with both opportunities and threats and has to do this in real time. This makes hardware incredibly important. Since we aren't computers except in an extremely metaphorical sense, hardware is all we've got, or at least, the hardware/software distinction in biology breaks down as a useful metaphor. And given that biological systems are also quite chaotic in the sense that minor perturbations can be amplified into large scale effects, the physical details of how our neural networks are implemented become important even at the molecular level. We're made of physical stuff, and nothing else, of course. But I think people tend to really under-estimate just how important what we're made of and how we're put together really is. We evolved from replicating molecule-like entities bit by bit, with each change built upon what came before, and at no point did biology just jettison the processes happening at the very lowest levels of processing. Functionalism and mind-as-computer are useful ideas and we shouldn't abandon them, but if you take the motivations of functionalism seriously, you have to apply them consistently all the way down, even to the kinds of exchanges happening between atoms and molecules within an in between neurons. It all counts. If functionalism is a useful paradigm for how the mind works, then I would fall back on the fact that computer software runs on machines whose molecular details are unimportant in the sense that an algorithm depends very much on a mathematical structure, and not at all on whether the computer uses NMOS or CMOS transistors. The semiconductor device physics certainly matters greatly concerning the operation of a single gate, but computers are assembled in such a way that the analog properties (those properties that might vary from one technology to another) are de-emphasized. That's an important part of the appeal of digital electronics. A binary adder is composed of a number of logic gates that produce the same result whether they're implemented with TTL or CMOS transistors, or even vacuum tubes. It seems the same is true of neural networks. A machine learning network trained to recognize pictures of hippos could be run on specialized hardware or in a software simulator. It might run faster on the specialized hardware but they produce the same result: the details of how the network nodes are implemented simply don't matter. |
|
