I'm struck with a similar impression to the one I had the first time I read the fading qualia thought experiment, that life seems easier for the functionalist, as you note in your post. For us, as soon as we're told that Entity-N is functionally isomorphic with Entity-0, the question is settled. It seems only by holding that experience is ontologically distinct from functionality that we end up with this tangled knot.
That said, a possible reply might be that you're essentially saying that a skeptic about Robot's consciousness can bite the bullet and just accept one of the horns Chalmers offers, or a hybrid between them. If Entity-0 is introspecting, but somewhere along the line Entity-J is only "introspecting", but still has a state causing it to express that experience remains, it seems instrumentally equivalent to just accepting that experience ends or fades somewhere along the line.
But I may well have missed a crucial point somewhere.
What you say seems right to me. It’s a kind of bullet biting move; but they were basically already biting that bullet, I think. So the argument makes a certain type on inelegance more vivid, but shouldn’t really move the target audience.
I take the target audience of Chalmers to be those who don’t buy into organizational invariance but who generally think our introspective faculties are reliable. For example, non-physicalists who want to rely on the reliability of introspection to make their case for non-physicalism, but who also aren’t partial to functional (of the nomological, not analytic, kind) accounts of consciousness that Chalmers endorses.
If that is so, then what Chalmers needs to show is that the unreliability of intermediate robot introspection infects and imperils our own introspective capacities. So your point that we should expect intermediate robot to have unreliable introspection on organizational invariance, while true, is not enough. You also have to show that it is simultaneously reasonable to expect this and reasonable to believe in the reliability of ordinary human introspection.
You do give the example of the red/green tree in the dancing qualia case, but I think the argument here is incomplete. The reason that our ordinary vision remains generally reliable even though it fails in the case of the red/green tree is precisely because such cases (and optical illusions in general) are extremely rare in the visual environment.
But why are we entitled to think that cases of fading/dancing robots are similarly rare in the physical world? After all, they are functionally the same as conscious subjects, so you can’t appeal to any evolutionary imperative which selected for a particular function of some sort. If dancing/fading robots were in fact in the minority, this would seem to simply be a matter of coincidence (e.g. that the psycho-physical laws were arrayed so and so…). And we have no reason to think this is the case.
Since we have no good defeater for thinking such cases would be common, we have no good defeater for the argument that unreliability of dancing/fading robot introspection infects the reliability of ordinary human introspection. And so your argument is seriously weakened I think. At the very least, it would be reasonable to worry that such an infection could take place, and that our introspective faculties might be compromised, in the absence of data as to the occurrence of weird changing qualia cases.
Having said all that, you could argue that it is reasonable to assume rarity of dancing qualia cases since it only pertains to the comparative phenomenal judgements of introspection, and not our regular phenomenal judgements.
Regular phenomenal judgements like “this pain is really bad” and “this color is blue” remain true I think for both faded (not fading) and inverted robot cases. If you’re the target audience here, then you don’t buy into organizational invariance, which presumably means you buy into a private account of phenomenal semantics and epistemics.
This would mean that statements like “this pain is excruciating” can be understood as comparative in reference to one’s own private phenomenal language. So even if my pain is very slight relative to yours, it’s still true that “I am in excruciating pain” because the pain is great compared to my other types of regular pain.
If this is true, then dancing/fading qualia robots really only make introspective errors when they are making comparative introspective judgements between their current and past experiences when they were not in a faded/inverted state. As long as dancing/fading qualia transitional cases were rare, your argument still goes through, even if static inverted/faded cases are common.
But I think the same worry still applies here. Yeah granted it would be really weird to think that there’s a serious possibility that we might be in dancing/fading qualia cases most of the time, but how could we ever have evidence for the rarity of such cases if we didn’t accept organizational invariance to begin with?
You do also bring up that fading/dancing robots might not even be introspecting at all in intermediate cases. That seems implausible to me on any general account of introspection. The fading qualia cases would have all the same phenomenal and cognitive tools that ordinary introspectors would have, except for the fact that their qualia is significantly less intense. But it seems weird to suggest that intensity of qualia should entirely make a difference as to whether you are capable of introspecting or not. We can introspect a minor pain (stubbing your toe) just as well as a major pain (being tortured).
Granted, there might be a point where experience becomes so faded that it might make introspection completely impossible, but we don’t need to invoke such cases for the argument to go through, it’s enough to say for example that a fading robot just experiences a dull ache where an ordinary person would experience an excruciating sensation of a broken bone etc…
The same applies for the dancing qualia, since it seems completely implausible to say that inverting your qualia would suddenly switch off your introspective capacities.
Thanks for this thoughtful comment, Alex! There’s a lot there. One thought I have is that rarity might not be the main issue. The target audience might think that humans are lucky to be made of conscious carbon in a world of mostly silicon zombies. Now if that’s not a reasonable view, fine, it might not be! But that’s the target audience that Fading Qualia needs to appeal to. On introspective capacities, I’m inclined to think that an important subset of the target audience would hold that without consciousness, there’s no real linguistic meaning or introspection at all. So if the qualia go, so does the introspection. (This wouldn’t apply in the Dancing case, though, so I didn’t mean to imply that introspection would be absent in that case.)
P.S. Also, I don’t think it’s unreasonable to hold that dancing/fading qualia cases are the majority of possible conscious observers if organizational invariance is false. After all, if nomological functionalism of the kind Chalmers advocates is false, then qualia wont be expected to track with functional states, so they must track some other thing.
But this other thing could be anything! From arbitrary stuff like individual neurons being composed in a particular way, to panpsychist views where even your teapot has complex qualia states. Our only limitation is that our qualia-physical matching laws have to be constructed in such a way to include us as conscious observers, but that still leaves an almost unlimited amount of possible pairings.
Given the vast possible array of physical-qualia matchings in the absence of organizational invariance, it would be a weird coincidence if physical-qualia matching states happened to align exactly with functional states.
And if the neural realizer states of particular conscious states don’t in fact track neatly with functional states, then as we proceed temporally from one functional state to another, we would reasonably expect our qualia to change. So if anything, we should expect dancing/fading qualia states to be in the majority of possible conscious experiences!
But since one of Chalmers stipulations is that introspective belief is tied to our functional states (see his reply to Searle), albeit perhaps grounded in consciousness, then it follows that almost everyone in our reference class will have dancing/fading qualia over time, which is a disaster for introspective reliability.
And sure you can just assert that you’re holding to a particular view of biological naturalism or something where this problem is mostly circumvented (i.e. one where the physical determinants of qualia happened to almost always track with functional states by chance), but the point is that we would have no introspective justification for that belief. If organizational invariance is false, our introspective reports are compatible with a bunch of possible physical-qualia matchings. So you can’t use this stipulation to hold that your introspection is reliable, since that’s begging the question.
I think the ultimate conclusion is that you have to go all the way. Either believe that both experience and introspection is non-functional (the Helen Yetter-Chappell view) or that both introspection and experience is functional. Chalmers argues in his original paper for the latter (of the nomological kind of course), although the other option is available too, and I’m not sure what his current views on the subject are now for that matter.
I hope this made sense! Thanks for your time Eric.
Haha yeah I wrote rather too much I think. I agree that many in the target audience might think that introspection is grounded in consciousness (and that’s a view I myself prefer), and that furthermore one could reasonably say that we are simply lucky minority conscious observers in a a world of mostly zombie functional isomorphs. But note that only works because of selection effects. If consciousness is needed for introspection, then obviously you won’t be a phenomenal zombie given that you are capable of introspecting. So it’s not a surprise that we find ourselves to be conscious observers, given that we’ve restricted our reference class in this way.
But note that this reply doesn’t work for cases of fading qualia observers who are clearly conscious (but whose qualia are still faded relative to observers like you and I). All I need you to grant is that there are such possible observers. Once this is granted, then it follows that rarity is still a factor. You have to assert that such observers (who are capable of introspection) still must be in the minority, otherwise it is legitimate for us to worry that our own introspective capacities are unreliable- because we belong in the same reference class (conscious observers) as fading qualia observers. My point is that we can’t justify our introspective reliability (or at least I haven’t seen an argument from you) unless we subscribe to functionalism.
So Chalmers point isn’t undermined, because we might still worry that in the absence of organizational invariance (and in the absence of an argument that fading/dancing qualia observers are in the minority) then we might be conscious fading/dancing qualia observers, which would entail introspective unreliability.
Notice also that the selection effect reply doesn’t apply here, since such observers are defined as being conscious.
An interesting argument Eric.
I'm struck with a similar impression to the one I had the first time I read the fading qualia thought experiment, that life seems easier for the functionalist, as you note in your post. For us, as soon as we're told that Entity-N is functionally isomorphic with Entity-0, the question is settled. It seems only by holding that experience is ontologically distinct from functionality that we end up with this tangled knot.
That said, a possible reply might be that you're essentially saying that a skeptic about Robot's consciousness can bite the bullet and just accept one of the horns Chalmers offers, or a hybrid between them. If Entity-0 is introspecting, but somewhere along the line Entity-J is only "introspecting", but still has a state causing it to express that experience remains, it seems instrumentally equivalent to just accepting that experience ends or fades somewhere along the line.
But I may well have missed a crucial point somewhere.
What you say seems right to me. It’s a kind of bullet biting move; but they were basically already biting that bullet, I think. So the argument makes a certain type on inelegance more vivid, but shouldn’t really move the target audience.
Hey Eric,
I take the target audience of Chalmers to be those who don’t buy into organizational invariance but who generally think our introspective faculties are reliable. For example, non-physicalists who want to rely on the reliability of introspection to make their case for non-physicalism, but who also aren’t partial to functional (of the nomological, not analytic, kind) accounts of consciousness that Chalmers endorses.
If that is so, then what Chalmers needs to show is that the unreliability of intermediate robot introspection infects and imperils our own introspective capacities. So your point that we should expect intermediate robot to have unreliable introspection on organizational invariance, while true, is not enough. You also have to show that it is simultaneously reasonable to expect this and reasonable to believe in the reliability of ordinary human introspection.
You do give the example of the red/green tree in the dancing qualia case, but I think the argument here is incomplete. The reason that our ordinary vision remains generally reliable even though it fails in the case of the red/green tree is precisely because such cases (and optical illusions in general) are extremely rare in the visual environment.
But why are we entitled to think that cases of fading/dancing robots are similarly rare in the physical world? After all, they are functionally the same as conscious subjects, so you can’t appeal to any evolutionary imperative which selected for a particular function of some sort. If dancing/fading robots were in fact in the minority, this would seem to simply be a matter of coincidence (e.g. that the psycho-physical laws were arrayed so and so…). And we have no reason to think this is the case.
Since we have no good defeater for thinking such cases would be common, we have no good defeater for the argument that unreliability of dancing/fading robot introspection infects the reliability of ordinary human introspection. And so your argument is seriously weakened I think. At the very least, it would be reasonable to worry that such an infection could take place, and that our introspective faculties might be compromised, in the absence of data as to the occurrence of weird changing qualia cases.
Having said all that, you could argue that it is reasonable to assume rarity of dancing qualia cases since it only pertains to the comparative phenomenal judgements of introspection, and not our regular phenomenal judgements.
Regular phenomenal judgements like “this pain is really bad” and “this color is blue” remain true I think for both faded (not fading) and inverted robot cases. If you’re the target audience here, then you don’t buy into organizational invariance, which presumably means you buy into a private account of phenomenal semantics and epistemics.
This would mean that statements like “this pain is excruciating” can be understood as comparative in reference to one’s own private phenomenal language. So even if my pain is very slight relative to yours, it’s still true that “I am in excruciating pain” because the pain is great compared to my other types of regular pain.
If this is true, then dancing/fading qualia robots really only make introspective errors when they are making comparative introspective judgements between their current and past experiences when they were not in a faded/inverted state. As long as dancing/fading qualia transitional cases were rare, your argument still goes through, even if static inverted/faded cases are common.
But I think the same worry still applies here. Yeah granted it would be really weird to think that there’s a serious possibility that we might be in dancing/fading qualia cases most of the time, but how could we ever have evidence for the rarity of such cases if we didn’t accept organizational invariance to begin with?
You do also bring up that fading/dancing robots might not even be introspecting at all in intermediate cases. That seems implausible to me on any general account of introspection. The fading qualia cases would have all the same phenomenal and cognitive tools that ordinary introspectors would have, except for the fact that their qualia is significantly less intense. But it seems weird to suggest that intensity of qualia should entirely make a difference as to whether you are capable of introspecting or not. We can introspect a minor pain (stubbing your toe) just as well as a major pain (being tortured).
Granted, there might be a point where experience becomes so faded that it might make introspection completely impossible, but we don’t need to invoke such cases for the argument to go through, it’s enough to say for example that a fading robot just experiences a dull ache where an ordinary person would experience an excruciating sensation of a broken bone etc…
The same applies for the dancing qualia, since it seems completely implausible to say that inverting your qualia would suddenly switch off your introspective capacities.
Thanks for reading all of this!
Thanks for this thoughtful comment, Alex! There’s a lot there. One thought I have is that rarity might not be the main issue. The target audience might think that humans are lucky to be made of conscious carbon in a world of mostly silicon zombies. Now if that’s not a reasonable view, fine, it might not be! But that’s the target audience that Fading Qualia needs to appeal to. On introspective capacities, I’m inclined to think that an important subset of the target audience would hold that without consciousness, there’s no real linguistic meaning or introspection at all. So if the qualia go, so does the introspection. (This wouldn’t apply in the Dancing case, though, so I didn’t mean to imply that introspection would be absent in that case.)
P.S. Also, I don’t think it’s unreasonable to hold that dancing/fading qualia cases are the majority of possible conscious observers if organizational invariance is false. After all, if nomological functionalism of the kind Chalmers advocates is false, then qualia wont be expected to track with functional states, so they must track some other thing.
But this other thing could be anything! From arbitrary stuff like individual neurons being composed in a particular way, to panpsychist views where even your teapot has complex qualia states. Our only limitation is that our qualia-physical matching laws have to be constructed in such a way to include us as conscious observers, but that still leaves an almost unlimited amount of possible pairings.
Given the vast possible array of physical-qualia matchings in the absence of organizational invariance, it would be a weird coincidence if physical-qualia matching states happened to align exactly with functional states.
And if the neural realizer states of particular conscious states don’t in fact track neatly with functional states, then as we proceed temporally from one functional state to another, we would reasonably expect our qualia to change. So if anything, we should expect dancing/fading qualia states to be in the majority of possible conscious experiences!
But since one of Chalmers stipulations is that introspective belief is tied to our functional states (see his reply to Searle), albeit perhaps grounded in consciousness, then it follows that almost everyone in our reference class will have dancing/fading qualia over time, which is a disaster for introspective reliability.
And sure you can just assert that you’re holding to a particular view of biological naturalism or something where this problem is mostly circumvented (i.e. one where the physical determinants of qualia happened to almost always track with functional states by chance), but the point is that we would have no introspective justification for that belief. If organizational invariance is false, our introspective reports are compatible with a bunch of possible physical-qualia matchings. So you can’t use this stipulation to hold that your introspection is reliable, since that’s begging the question.
I think the ultimate conclusion is that you have to go all the way. Either believe that both experience and introspection is non-functional (the Helen Yetter-Chappell view) or that both introspection and experience is functional. Chalmers argues in his original paper for the latter (of the nomological kind of course), although the other option is available too, and I’m not sure what his current views on the subject are now for that matter.
I hope this made sense! Thanks for your time Eric.
Haha yeah I wrote rather too much I think. I agree that many in the target audience might think that introspection is grounded in consciousness (and that’s a view I myself prefer), and that furthermore one could reasonably say that we are simply lucky minority conscious observers in a a world of mostly zombie functional isomorphs. But note that only works because of selection effects. If consciousness is needed for introspection, then obviously you won’t be a phenomenal zombie given that you are capable of introspecting. So it’s not a surprise that we find ourselves to be conscious observers, given that we’ve restricted our reference class in this way.
But note that this reply doesn’t work for cases of fading qualia observers who are clearly conscious (but whose qualia are still faded relative to observers like you and I). All I need you to grant is that there are such possible observers. Once this is granted, then it follows that rarity is still a factor. You have to assert that such observers (who are capable of introspection) still must be in the minority, otherwise it is legitimate for us to worry that our own introspective capacities are unreliable- because we belong in the same reference class (conscious observers) as fading qualia observers. My point is that we can’t justify our introspective reliability (or at least I haven’t seen an argument from you) unless we subscribe to functionalism.
So Chalmers point isn’t undermined, because we might still worry that in the absence of organizational invariance (and in the absence of an argument that fading/dancing qualia observers are in the minority) then we might be conscious fading/dancing qualia observers, which would entail introspective unreliability.
Notice also that the selection effect reply doesn’t apply here, since such observers are defined as being conscious.