6 ms·

I fail to observe the difference between simulating a human mind on a computer and on a synaptic network. Why would a program run differently depending on the u

by apk-d 9y ago

I fail to observe the difference between simulating a human mind on a computer and on a synaptic network. Why would a program run differently depending on the underlying hardware? (assuming sufficiently faithful simulation as opposed to some approximation)

erokar 9y ago
Searle's argument is that consiousness is not a function of the program (there is no "program" in the human mind), but a function of hardware. Though just how the meat machine that is our brains produces consiousness is very poorly understood, it stand to reason that a computer is so different from it physically that it's no more likely a computer can acheive consoiusness than a stone or a car.
- apk-d 9y ago
  To me that sounds more like an immediate presupposition than an argument. I don't see why I'd favor an explanation of "consciousness is computation within a special magic meat brain" over simply "consciousness is computation", at least until we determine that brains run on special meat magic, of which so far there appears to be little evidence. I'm also not entirely convinced by implicit promises that "we'll discover this later when our understanding of brains is better".
  - foldr 9y ago
    Searle's point is that given what we currently understand about consciousness (i.e. nothing), there is no reason at all to think that consciousness will arise in any physical system that implements a particular kind of program. You talk about "magic meat brains", but one could just as well talk about "magic programs", since we have no idea how implementing a particular set of computations could give rise to conscious experience. Searle's bet is that the specific physical properties of the brain will turn out to be important for consciousness. Can we be sure of that? Of course not. But everyone in this domain is guessing.
    - hackinthebochs 9y ago
      > specific physical properties of the brain will turn out to be important for consciousness. But we do know something about the components of brains and they don't seem to exhibit any features that cannot in principle be computed. >But everyone in this domain is guessing. Sure we're guessing, but Searle goes further to rule out potential solutions based on assumptions of specialness of brains.
      - foldr 9y ago
        >But we do know something about the components of brains and they don't seem to exhibit any features that cannot in principle be computed. Sure, but so what? Neither do rain storms, but as Searle puts it, no-one would expect a computer simulation of a rain storm to actually make anyone wet. If conscious brains can be simulated, that in no way entails that simulations of conscious brains are themselves conscious. It seems to be a common misreading of Searle that he thinks that brains can’t be simulated on computers. The starting point of the Chinese room argument is to concede for the sake of argument that this is possible.
        hackinthebochs 9y ago
        >Neither do rain storms, but as Searle puts it, no-one would expect a computer simulation of a rain storm to actually make anyone wet. Of course not, but that's because the simulation doesn't have the right kind of causal relationship with my head to cause my head to get wet. But it might have the right kind of causal relationship with a simulated head to cause it to get wet. The analogy doesn't hold in the case of consciousness because all that is at stake is its own access to phenomenal properties. There is no causal category mismatch like there is with simulated rain and my physical head. If phenomenal properties are purely relational or functional, then the right kind of simulation will be conscious.
        foldr 9y ago
        It’s virtually a contradiction in terms to say that a phenomenal property is purely relational or functional. If it were obvious, or even plausible, that phenomenal properties were purely relational or functional, then no-one would be worrying about their implications for the philosophy of mind!
- cygx 9y ago
  But hardware can be simulated in software.
cgmg 9y ago
This comment is relevant: https://www.reddit.com/r/artificial/comments/5hmduk/prof_schmidhuber_the_problems_of_ai_consciousness/db23i91/ https://www.reddit.com/r/artificial/comments/5hmduk/prof_sch.... > For example, nuclear energy is a real physical phenonenon. It doesn't exist because of various abstract relations - i.e., simulating a nuclear reactor in a computer doesn't mean you have nuclear energy. We know that matter of a specific kind arranged in a specific way creates nuclear energy. > do strong AI proponents think that causal relationships must be involved to run a program to make it conscious? Or is it enough for the abstract relationships to exist? For example, how about a computer program written down on a piece of paper? Yes or no? Why is the physical running of it important? If so, please explain the physics of how running it in dominoes, water valves or transistors all produce the same phenomenon. If not, does this mean that any abstract set of relations is also conscious - the program on a piece of paper? Doesn't that then also mean that there are an infinitude of consciousnesses since an infinitude of abstract relations exist between all of the bits of matter in the universe? Analogy: Simulating fluid dynamics on a computer does not mean the computer becomes wet. Simulating a black hole on a computer does not mean the computer starts curving the spacetime around it. Simulating an electric field on a computer does not mean the computer creates an electric field. Simulating a brain on a computer may or may not mean the computer creates consciousness.

by apk-d 9y ago

erokar 9y ago

Searle's argument is that consiousness is not a function of the program (there is no "program" in the human mind), but a function of hardware. Though just how the meat machine that is our brains produces consiousness is very poorly understood, it stand to reason that a computer is so different from it physically that it's no more likely a computer can acheive consoiusness than a stone or a car.

apk-d 9y ago
To me that sounds more like an immediate presupposition than an argument. I don't see why I'd favor an explanation of "consciousness is computation within a special magic meat brain" over simply "consciousness is computation", at least until we determine that brains run on special meat magic, of which so far there appears to be little evidence. I'm also not entirely convinced by implicit promises that "we'll discover this later when our understanding of brains is better".
- foldr 9y ago
  Searle's point is that given what we currently understand about consciousness (i.e. nothing), there is no reason at all to think that consciousness will arise in any physical system that implements a particular kind of program. You talk about "magic meat brains", but one could just as well talk about "magic programs", since we have no idea how implementing a particular set of computations could give rise to conscious experience. Searle's bet is that the specific physical properties of the brain will turn out to be important for consciousness. Can we be sure of that? Of course not. But everyone in this domain is guessing.
  - hackinthebochs 9y ago
    > specific physical properties of the brain will turn out to be important for consciousness. But we do know something about the components of brains and they don't seem to exhibit any features that cannot in principle be computed. >But everyone in this domain is guessing. Sure we're guessing, but Searle goes further to rule out potential solutions based on assumptions of specialness of brains.
    - foldr 9y ago
      >But we do know something about the components of brains and they don't seem to exhibit any features that cannot in principle be computed. Sure, but so what? Neither do rain storms, but as Searle puts it, no-one would expect a computer simulation of a rain storm to actually make anyone wet. If conscious brains can be simulated, that in no way entails that simulations of conscious brains are themselves conscious. It seems to be a common misreading of Searle that he thinks that brains can’t be simulated on computers. The starting point of the Chinese room argument is to concede for the sake of argument that this is possible.
      - hackinthebochs 9y ago
        >Neither do rain storms, but as Searle puts it, no-one would expect a computer simulation of a rain storm to actually make anyone wet. Of course not, but that's because the simulation doesn't have the right kind of causal relationship with my head to cause my head to get wet. But it might have the right kind of causal relationship with a simulated head to cause it to get wet. The analogy doesn't hold in the case of consciousness because all that is at stake is its own access to phenomenal properties. There is no causal category mismatch like there is with simulated rain and my physical head. If phenomenal properties are purely relational or functional, then the right kind of simulation will be conscious.
        foldr 9y ago
        It’s virtually a contradiction in terms to say that a phenomenal property is purely relational or functional. If it were obvious, or even plausible, that phenomenal properties were purely relational or functional, then no-one would be worrying about their implications for the philosophy of mind!
cygx 9y ago
But hardware can be simulated in software.

cgmg 9y ago

This comment is relevant: https://www.reddit.com/r/artificial/comments/5hmduk/prof_schmidhuber_the_problems_of_ai_consciousness/db23i91/ https://www.reddit.com/r/artificial/comments/5hmduk/prof_sch.... > For example, nuclear energy is a real physical phenonenon. It doesn't exist because of various abstract relations - i.e., simulating a nuclear reactor in a computer doesn't mean you have nuclear energy. We know that matter of a specific kind arranged in a specific way creates nuclear energy. > do strong AI proponents think that causal relationships must be involved to run a program to make it conscious? Or is it enough for the abstract relationships to exist? For example, how about a computer program written down on a piece of paper? Yes or no? Why is the physical running of it important? If so, please explain the physics of how running it in dominoes, water valves or transistors all produce the same phenomenon. If not, does this mean that any abstract set of relations is also conscious - the program on a piece of paper? Doesn't that then also mean that there are an infinitude of consciousnesses since an infinitude of abstract relations exist between all of the bits of matter in the universe? Analogy: Simulating fluid dynamics on a computer does not mean the computer becomes wet. Simulating a black hole on a computer does not mean the computer starts curving the spacetime around it. Simulating an electric field on a computer does not mean the computer creates an electric field. Simulating a brain on a computer may or may not mean the computer creates consciousness.