The Plant Consciousness Wars

Does it? I've never read the proof.

I think I probably am missing the point of the argument. I was only ever an applied mathematician, not a pure one. But evidently the argument has been severely criticised by those who do understand it.

I meant all proofs have some involvement of intuition, and what exactly is the grounding of a proof but the shared intuitive sense of Logic's authority?

The Godelian argument has its detractors and defenders, though I don't think it is necessary to show consciousness cannot be a computer program [if Physicalism is true].

Which is different than say an android, which I would think could be conscious on metaphysically neutral grounds.

[Laird]

Are you essentially saying that any kind of consciousness arising solely out of physical mechanisms would have to be epiphenomenal, in the sense that it could be separated from the physical mechanisms and would have no influence on them?

Something close to that, yes. I'm saying that this artificial brain would deterministically compute in the same way whether or not it became conscious. Any consciousness, then, could only "float off the top of" the computations and could not influence them.

And therefore it's an impossibility?

The impossibility doesn't come from the epiphenomenalism as such, but from the stipulation that this artificial brain behaves like natural humans do: that is, that it reflects on (and expresses descriptions of) its own consciousness. This is where Titus's argument comes in. Essentially, the argument runs (and hopefully I paraphrase it reasonably) that this is impossible because that reflection and expression occurs in the physical calculations, which (can) have no knowledge of consciousness because they themselves are non-conscious and because consciousness has no effect on them which could "inform" them as to what consciousness is/means.

It feels to me as though consciousness should be indissolubly bound up with the processing of the information, and therefore that it can't be said that consciousness has no influence on the physical mechanisms.

But what does being indissolubly bound up mean in practice? Unless it means that the processing is changed in some way [edit: from what it was / would have been in the absence of consciousness], I can't see how it [consciousness] can meaningfully be said to "influence" that processing.

Something close to that, yes. I'm saying that this artificial brain would deterministically compute in the same way whether or not it became conscious. Any consciousness, then, could only "float off the top of" the computations and could not influence them.

What feels wrong to me is the separation of consciousness from computation. If the computation is producing the consciousness, I don't think it makes sense that the computation could be identical in two cases - one with and one without consciousness. I think if the computation were the same, the consciousness would necessarily also be the same.

There is a fascinating article on the science of this controversy in Trends in Plant Science. One idea on how to get around the lack in plants of any of the neurological structures known to be necessary for consciousness in animals is "swarm intelligence", but it is attacked as having a lot of problems.

Thanks for the article though it just seems very odd that we could even be in the position to deny Consciousness to plants, when we cannot even account for why we have it.

I'd be curious what mystical experience has said about plant consciousness, though there might not be enough references to plant mentality to be persuasive even for the mystic-friendly proponent...

[Laird]

What feels wrong to me is the separation of consciousness from computation. If the computation is producing the consciousness, I don't think it makes sense that the computation could be identical in two cases - one with and one without consciousness. I think if the computation were the same, the consciousness would necessarily also be the same.

I think you're saying that a given computation will or will not produce consciousness, and thus hypotheticals which assume that that computation could be either producing consciousness or not producing it are invalid.

That's a reasonable position to take, however, it doesn't change my point: the calculations are deterministic; there is in any case no way for consciousness to influence them.

I'm not sure whether you're objecting to my point, but in case you are, here's a somewhat intuitive argument which I hope gets the point across better than simply asserting it does, even though I do think that the assertion is self-evident.

Rather than looking at a given artificial brain and considering it with or without consciousness, which you reasonably point out might not be coherent, we could look at two artificial brains of slightly different degrees of complexity. The first is just below the degree of complexity (similarity to a human brain) to entail consciousness; the second is just above it; other than that the two are identical. Let's say that the difference is a single line of code which translates into a single instruction in the processor on which the brain is running (the argument wouldn't change if it were multiple lines of code, it would just be less stark).

Now, on the view you're expressing, consciousness would be indissolubly bound up with [edit: in the sense of being able to influence processing] that single instruction [alone], because consciousness wasn't present in the slightly less complex artificial brain; this is the only place where it could have an effect. But clearly to say that that single instruction "influences" the artificial brain in any way that could be attributed to consciousness is mistaken - it is simply an instruction carried out by a CPU in accordance with the rules of that CPU. Any consciousness with which it is "bound up" can only "come along for the ride"; it is purely an epiphenomenon.

I think you're saying that a given computation will or will not produce consciousness, and thus hypotheticals which assume that that computation could be either producing consciousness or not producing it are invalid.

That's a reasonable position to take, however, it doesn't change my point: the calculations are deterministic; there is in any case no way for consciousness to influence them.

I'm not sure whether you're objecting to my point, but in case you are, here's a somewhat intuitive argument which I hope gets the point across better than simply asserting it does, even though I do think that the assertion is self-evident.

Rather than looking at a given artificial brain and considering it with or without consciousness, which you reasonably point out might not be coherent, we could look at two artificial brains of slightly different degrees of complexity. The first is just below the degree of complexity (similarity to a human brain) to entail consciousness; the second is just above it; other than that the two are identical. Let's say that the difference is a single line of code which translates into a single instruction in the processor on which the brain is running (the argument wouldn't change if it were multiple lines of code, it would just be less stark).

Now, on the view you're expressing, consciousness would be indissolubly bound up with [edit: in the sense of being able to influence processing] that single instruction [alone], because consciousness wasn't present in the slightly less complex artificial brain; this is the only place where it could have an effect. But clearly to say that that single instruction "influences" the artificial brain in any way that could be attributed to consciousness is mistaken - it is simply an instruction carried out by a CPU in accordance with the rules of that CPU. Any consciousness with which it is "bound up" can only "come along for the ride"; it is purely an epiphenomenon.

I think in the second paragraph your argument still depends on making a separation between the calculation and the consciousness, which is what feels wrong to me. After all, there's no doubt that past calculations can influence future calculations. So if calculation couldn't be separated from consciousness, it wouldn't be correct to say that consciousness couldn't influence calculations.

Regarding the calculations being deterministic, I wasn't necessarily assuming that. And if they're not, and if consciousness can influence random processes, how does that affect the argument? (Come to think of it, can we rule out consciousness influencing even deterministic processes?) That is maybe a side issue, but if we're open to any kind of PK perhaps it should be considered.

[Laird]

I think in the second paragraph your argument still depends on making a separation between the calculation and the consciousness, which is what feels wrong to me. After all, there's no doubt that past calculations can influence future calculations. So if calculation couldn't be separated from consciousness, it wouldn't be correct to say that consciousness couldn't influence calculations.

I'm not sure what you mean by "calculation couldn't be separated from consciousness". You seem to be trying to avoid epiphenominalism and its implications, so then you must be arguing for some sort of identity theory? That in some sense calculation "is" consciousness, and vice versa? Or... what?

Regarding the calculations being deterministic, I wasn't necessarily assuming that. And if they're not, and if consciousness can influence random processes, how does that affect the argument?

Well, if you're going for some sort of identity theory, then consciousness influencing random processes amounts to calculations influencing random processes, which is hard to understand. You'd seem to then be arguing something like: a series of instructions on a CPU (which are indissolubly bound with, and perhaps in some sense identical with, consciousness) influence the hardware true random generator on that same CPU (or on some other connected hardware device)?

I think in the second paragraph your argument still depends on making a separation between the calculation and the consciousness, which is what feels wrong to me. After all, there's no doubt that past calculations can influence future calculations. So if calculation couldn't be separated from consciousness, it wouldn't be correct to say that consciousness couldn't influence calculations.

Curious what calculations are in this sense, though probably also need to clarify "information" and "processing"?

To tie this back to plants, is the potential exploitation of superposition in photosynthesis a calculation or just something that happens? Of course, similarly, are any of our biological reactions to our environments calculations, information processing, or simply events that tie into our sensations/qualia?

It seems to me that to accept an AI as conscious might possibly also be an argument for plant consciousness? Can you have the former without making an argument for the latter?

[Laird]

Regarding the calculations being deterministic, I wasn't necessarily assuming that. And if they're not, and if consciousness can influence random processes, how does that affect the argument?

Well, if you're going for some sort of identity theory, then consciousness influencing random processes amounts to calculations influencing random processes, which is hard to understand. You'd seem to then be arguing something like: a series of instructions on a CPU (which are indissolubly bound with, and perhaps in some sense identical with, consciousness) influence the hardware true random generator on that same CPU (or on some other connected hardware device)?

But if we stick to a dualist theory, then, yes, consciousness being able to influence the random processes with which it is associated could help to avoid the epiphenomenalist conclusion.

But if we stick to a dualist theory, then, yes, consciousness being able to influence the random processes with which it is associated could help to avoid the epiphenomenalist conclusion.

Do plants have souls as well as bodies?