4 ms·

Why would computational explanations of consciousness be more ridiculous than the non-explanation that is asserting everything is conscious in some way (which i

Why would computational explanations of consciousness be more ridiculous than the non-explanation that is asserting everything is conscious in some way (which is basically what panpsychism means)?

ethanbond 3y ago
At what degree of computation does consciousness emerge? There must be some level of computation at which there’s zero consciousness and then one rung above that there’s a non-zero degree of consciousness. Where could that line possibly be? I personally cannot reasonably even guess at such a boundary, so panpsychism is a much cleaner explanation. These two ideas aren’t really mutually exclusive anyway. I.e. you can believe that consciousness is what it feels like to compute, but then “what is compute” except a particularly structured form of a thing that’s happening everywhere and all the time (particles and energy bouncing around) in more or less structured forms. And if you say “well not all energy/particle interactions are computation, even of an abysmally low degree,” then return to the first question: what types of interactions or what degrees of structure are non-computational where one rung above it they are computational?
- simiones 3y ago
  The computational explanation would hold that there is some specific computation that can be identified as what we call consciousness. This would be equivalent to how we say that a specific computation is what we call "sorting". It's not a gradualist notion. A computation either is or isn't consciousness. If it is, then an agent which is currently performing that computation is conscious. If an agent is performing a computation that is not consciousness is not conscious. Similarly, certain computations represent a sort. Agents which are performing that computation are sorting. The clearest case is a computer running a sort algorithm, but there are also purely physical processes that perform the same computations, and animals or people that do it - so "agent" here is a very broad term. Agents which are performing a computation that is not a sort are not sorting. Of course, determining which computation is actually consciousness is a very difficult problem, well beyond our current knowledge. As such, it is also obvious that it may turn out that this is not a correct model. But I don't see why this would be considered a ridiculous model of what consciousness means. It is also a model which can, in principle, be falsified, and it could actually help resolve certain difficult real-world problems. For example, if we did have a satisfying definition of which computations represent consciousness, we could then use it to determine at which point a child becomes conscious, or whether a particular animal is conscious or not (assuming we consider that conscious === should not be killed arbitrarily, which is a completely different debate). By contrast, panpsychism doesn't really resolve anything and can't be used to determine the answer to this type of moral question even in principle (if any system is somewhat conscious, then even swiveling my chair means that I am destroying a conscious system; but also creating another one).
  - ethanbond 3y ago
    A “computation” is a concept that we have in our heads. Computations only actually occur as an interaction among some substrate of the physical universe. So you’re still saying that if you arrange atoms and energy in the right configuration then we go from zero consciousness to some consciousness. If you can find that configuration I’m certainly open to being convinced. Even your “non-gradualist” explanation does in fact run into gradualist problems, as you mention one of when a child “becomes conscious.” You assert that this theory is falsifiable: how so? You can’t even falsify the notion that there’s only one consciousness at all in the entire universe (yours, the one you’re “doing”). Regarding moral implications, inconvenient moral conclusions are not the same as “no help in arriving at moral conclusions.” You just named one such conclusion that you’d be compelled to arrive at (I wouldn’t, personally), so it’s not really true it doesn’t help determine the answer.
    - simiones 3y ago
      > A “computation” is a concept that we have in our heads. Computations only actually occur as an interaction among some substrate of the physical universe. So you’re still saying that if you arrange atoms and energy in the right configuration then we go from zero consciousness to some consciousness. If you can find that configuration I’m certainly open to being convinced. A computation is a physical process. A pendulum swinging is a computation of the path that the pendulum will take (and of many other things). Whether there is some conscious being that observes and understands that computation or not is irrelevant to whether the computation is happening (that is, it won't affect the result in any way). And yes, I am saying that one possible theory of consciousness is that some particular arrangement of matter and energy evolving in a certain way over time is consciousness. That is what I mean by a computational theory of consciousness. I am not in any way claiming I could define today which particular arrangement and evolution that is - I explicitly said I believe the specific model is far beyond our current understanding of the subject. > Even your “non-gradualist” explanation does in fact run into gradualist problems, as you mention one of when a child “becomes conscious.” I wouldn't call a theory that tells you that a system is either conscious or unconscious "gradualist". Yes, the system is not conscious at time A and becomes conscious at time A+1. This happens exactly the same way as any other computation: my processor is currently not performing a sort, and when it reaches the sort() call, it becomes sorting all of a sudden; when the call to sort() finishes, it is no longer sorting. Similarly, we can imagine a procedure consciousness(); when a system starts executing this procedure, by definition it becomes conscious; when it finishes executing it, it becomes not conscious. > You assert that this theory is falsifiable: how so? You can’t even falsify the notion that there’s more than one consciousness at all in the entire universe (yours, the one you’re “doing”). Well, it is falsifiable in two ways. First, there is the same extent as any mathematical model is falsifiable: we can check whether it accurately accounts for our intuitions. It's the same way we can check whether the formal definition of a continuous function is correct. Now, once we accept that the proposed model matches what we intuitively mean by consciousness, it is further just as falsifiable as any other physical theory - it makes it so that the question "is system A conscious" (within the formal definition of consciousness I defined above) becomes a measurable claim about whether a particular system is performing a particular computation. > Regarding moral implications, inconvenient moral conclusions are not the same as “no help in arriving at moral conclusions.” You just named one such conclusion that you’d be compelled to arrive at (I wouldn’t, personally), so it’s not really true it doesn’t help determine the answer. If the conclusion is "any act you make is just as morally evil, since it leads to destroying a conscious system", that's not just inconvenient, it is by definition not helpful in making moral choices. It is as useless for arriving at moral conclusions as a very convenient model which says "no entity is conscious, so any choice you make is morally good". Alternatively, if you believe the consciousness of a subject is irrelevant to moral choices, then any model of consciousness is also not helpful in arriving at moral conclusions. And, if you believe that the amount of consciousness is the relevant bit for moral choices about a system, than still panpsychism is no help, since it only asserts that all things are conscious, it doesn't promise a model for determining which are more conscious and which are less conscious (you would need an additional model on top of panpsychism, perhaps something like integrated information theory at least).

Why would computational explanations of consciousness be more ridiculous than the non-explanation that is asserting everything is conscious in some way (which is basically what panpsychism means)?

ethanbond 3y ago

At what degree of computation does consciousness emerge? There must be some level of computation at which there’s zero consciousness and then one rung above that there’s a non-zero degree of consciousness. Where could that line possibly be? I personally cannot reasonably even guess at such a boundary, so panpsychism is a much cleaner explanation. These two ideas aren’t really mutually exclusive anyway. I.e. you can believe that consciousness is what it feels like to compute, but then “what is compute” except a particularly structured form of a thing that’s happening everywhere and all the time (particles and energy bouncing around) in more or less structured forms. And if you say “well not all energy/particle interactions are computation, even of an abysmally low degree,” then return to the first question: what types of interactions or what degrees of structure are non-computational where one rung above it they are computational?

simiones 3y ago
The computational explanation would hold that there is some specific computation that can be identified as what we call consciousness. This would be equivalent to how we say that a specific computation is what we call "sorting". It's not a gradualist notion. A computation either is or isn't consciousness. If it is, then an agent which is currently performing that computation is conscious. If an agent is performing a computation that is not consciousness is not conscious. Similarly, certain computations represent a sort. Agents which are performing that computation are sorting. The clearest case is a computer running a sort algorithm, but there are also purely physical processes that perform the same computations, and animals or people that do it - so "agent" here is a very broad term. Agents which are performing a computation that is not a sort are not sorting. Of course, determining which computation is actually consciousness is a very difficult problem, well beyond our current knowledge. As such, it is also obvious that it may turn out that this is not a correct model. But I don't see why this would be considered a ridiculous model of what consciousness means. It is also a model which can, in principle, be falsified, and it could actually help resolve certain difficult real-world problems. For example, if we did have a satisfying definition of which computations represent consciousness, we could then use it to determine at which point a child becomes conscious, or whether a particular animal is conscious or not (assuming we consider that conscious === should not be killed arbitrarily, which is a completely different debate). By contrast, panpsychism doesn't really resolve anything and can't be used to determine the answer to this type of moral question even in principle (if any system is somewhat conscious, then even swiveling my chair means that I am destroying a conscious system; but also creating another one).
- ethanbond 3y ago
  A “computation” is a concept that we have in our heads. Computations only actually occur as an interaction among some substrate of the physical universe. So you’re still saying that if you arrange atoms and energy in the right configuration then we go from zero consciousness to some consciousness. If you can find that configuration I’m certainly open to being convinced. Even your “non-gradualist” explanation does in fact run into gradualist problems, as you mention one of when a child “becomes conscious.” You assert that this theory is falsifiable: how so? You can’t even falsify the notion that there’s only one consciousness at all in the entire universe (yours, the one you’re “doing”). Regarding moral implications, inconvenient moral conclusions are not the same as “no help in arriving at moral conclusions.” You just named one such conclusion that you’d be compelled to arrive at (I wouldn’t, personally), so it’s not really true it doesn’t help determine the answer.
  - simiones 3y ago
    > A “computation” is a concept that we have in our heads. Computations only actually occur as an interaction among some substrate of the physical universe. So you’re still saying that if you arrange atoms and energy in the right configuration then we go from zero consciousness to some consciousness. If you can find that configuration I’m certainly open to being convinced. A computation is a physical process. A pendulum swinging is a computation of the path that the pendulum will take (and of many other things). Whether there is some conscious being that observes and understands that computation or not is irrelevant to whether the computation is happening (that is, it won't affect the result in any way). And yes, I am saying that one possible theory of consciousness is that some particular arrangement of matter and energy evolving in a certain way over time is consciousness. That is what I mean by a computational theory of consciousness. I am not in any way claiming I could define today which particular arrangement and evolution that is - I explicitly said I believe the specific model is far beyond our current understanding of the subject. > Even your “non-gradualist” explanation does in fact run into gradualist problems, as you mention one of when a child “becomes conscious.” I wouldn't call a theory that tells you that a system is either conscious or unconscious "gradualist". Yes, the system is not conscious at time A and becomes conscious at time A+1. This happens exactly the same way as any other computation: my processor is currently not performing a sort, and when it reaches the sort() call, it becomes sorting all of a sudden; when the call to sort() finishes, it is no longer sorting. Similarly, we can imagine a procedure consciousness(); when a system starts executing this procedure, by definition it becomes conscious; when it finishes executing it, it becomes not conscious. > You assert that this theory is falsifiable: how so? You can’t even falsify the notion that there’s more than one consciousness at all in the entire universe (yours, the one you’re “doing”). Well, it is falsifiable in two ways. First, there is the same extent as any mathematical model is falsifiable: we can check whether it accurately accounts for our intuitions. It's the same way we can check whether the formal definition of a continuous function is correct. Now, once we accept that the proposed model matches what we intuitively mean by consciousness, it is further just as falsifiable as any other physical theory - it makes it so that the question "is system A conscious" (within the formal definition of consciousness I defined above) becomes a measurable claim about whether a particular system is performing a particular computation. > Regarding moral implications, inconvenient moral conclusions are not the same as “no help in arriving at moral conclusions.” You just named one such conclusion that you’d be compelled to arrive at (I wouldn’t, personally), so it’s not really true it doesn’t help determine the answer. If the conclusion is "any act you make is just as morally evil, since it leads to destroying a conscious system", that's not just inconvenient, it is by definition not helpful in making moral choices. It is as useless for arriving at moral conclusions as a very convenient model which says "no entity is conscious, so any choice you make is morally good". Alternatively, if you believe the consciousness of a subject is irrelevant to moral choices, then any model of consciousness is also not helpful in arriving at moral conclusions. And, if you believe that the amount of consciousness is the relevant bit for moral choices about a system, than still panpsychism is no help, since it only asserts that all things are conscious, it doesn't promise a model for determining which are more conscious and which are less conscious (you would need an additional model on top of panpsychism, perhaps something like integrated information theory at least).