What Panpsychists Should Reject: On the Incompatibility of Panpsychism and Organizational Invariantism Presenter: Miguel Sebastian, National Autonomous University of Mexico Get Miguel’s paper Commenter 1: Hedda Hassel, University of Oslo Get Hedda comments Commenter 2: Philip Goff, University of Liverpool Get Philip’s comments Share this:FacebookTwitterEmailRedditPrintPinterestTumblrLinkedInLike this:Like Loading... Related 43 Comments hi miguel — nice paper. i agree that there is at least a tension here. prima facie organizational invariance (POI) suggests that consciousness is determined by organization alone, while constitutive panpsychism (CPP) suggests that it is determined by micro constituents, perhaps in conjunction with relations among them. i agree with hedda that non-constitutive panpsychism doesn’t suffer from the tension, but the constitutive version of panpsychism is the one that i’m most interested in (since it has the potential to avoid the twin threats of epiphenomenalism and interfering with causal closure), so i’ll focus on that version. a few notes: (1) as i say in “the character of consciousness” (p. 24), i’m now inclined to think that the dancing qualia argument is not as strong as the fading qualia argument. on my panpsychist days i’m particularly sympathetic with rejecting the former. if the fading qualia arguments works but the dancing qualia argument doesn’t, then the property of being conscious is an organizational invariant, but specific phenomenal properties may not be. (hedda in effect points toward a view like this in her last section.) this weaker version of POI can allow that differences in the microphenomenal constituents make a difference to the qualitative character of an experience (phenomenal red vs phenomenal green, say). they won’t make a difference to whether the constituted system is conscious, but that’s not so surprising, as all of them are phenomenal properties. (2) this weakened line is also congenial to the line you discuss where organization determines structure but not intrinsic character. i didn’t understand your response that this requires distinct “supersimilar” experiences that can’t be told apart experientially. phenomenally red and phenomenally green experiences aren’t supersimilar, as they can be told apart experientially. to be sure, the dancing qualia subject can’t tell them apart, but that’s an unusual subject in a highly unusual situation. (3) despite my attraction to this weakened version of POI, i don’t think that the full version is truly incompatible with CPP. to reconcile them, one just needs to say that macrophenomenal states are determined by microphenomenal constituents along with their organization, and that the same macrophenomenal state can be multiply realized by many different microphenomenal states. (your argument at the end of 3.1 seems to overlook the possibility of this sort of multiple realizability.) on this view the microphenomenology can still play a constitutive role, just a multiply realized one, and one on which the specific differences between microphenomenal states don’t matter much. one version of this view would be one that distinguishes subjective character and qualitative character, holding that microphenomenal states constitutively constitute subjective character by contributing the property of being conscious, while it’s the organization that’s responsible for the qualitative character. this would perhaps be an unusual view for a constitutive panpsychist, but i don’t think it’s incoherent. (4) your second argument for incompatibility, in 3.2, is basically a version of the dancing qualia argument run in a nomologically possible but logically possible world. i’ve always allowed that dancing qualia are at least logically possible — i think i can conceive of them, they’re just very odd. so i don’t think that dancing qualia arguments in nomologically impossible worlds get much purchase. Hi Dave, thank you very much for your notes. I will come back to them immediately after this post. First I would like to thank Hedda and Philip for their thoughtful and useful comments. I will try to rejoin some of the objections they present. Let me start with Hedda’s comments (I hope I don’t miss any important point. If so, please bring it back in the discussion) In the first section of her comments Hedda concedes that my argument provides additional reasons to give up on the claim that proto-phenomenal properties are constitutive of consciousness but that the problem was already there (the combination problem), so she is happy giving up constitutive panpsychism. I think that things are not so easy because constitution is what provides a reply to the modal argument. Proto-phenomenal properties constitute consciouness and therefore, althought there are worlds that verify P and not Q there is no world that that satisfy P and also not Q. It should be noted that explanation can be left out of the picture. As Hedda notes, quoting Philip, there is an explanatory gap between proto-phenomenal properties and phenomenal properties but panpsychism provides an explanation of why this is the case and this is enough for blocking the argument. I think that Hedda is right that giving up constitution blocks my first objection but it also loses the motivation from the conceivability argument. As I intended to make it clear in the paper the target of the paper is to show the tension between the principles that back up the endorsement of panpsychism due to the conceivability argument and the ones that back up the fading and dancing qualia. In section two, Hedda focuses on what I call the second pass of the argument. Hedda present two ways of blocking the argument. In the second one she claims that suddenly disappearing qualia are not problematic in the case of panspychism. I do not see much of a difference with the ordinary case but before entering the discussion in this point it might help if you could please present an answer along these lines to the dancing qualia case that I present. In the first one, she suggests that one might deny that the scenario that I describe is metaphysically possible by denying that strings and string- can interact. Such an interaction would require “the interaction of substances with radically different intrinsic natures, and would be faced with classic interaction problems of substance dualism”. I do not see why such interaction would be problematic. If our Physics go in the right direction P is independent of the intrinsic nature of the fundamental entities. But they might be wrong in this and Hedda suggests the possibility that”causal relations are entered into in virtue of the intrinsic properties of the relata, or that dispositions are categorically grounded.”. This is already a step far beyond the claim that consciousness depends on the intrinsic nature of the fundamental entities of the actual world that is what the conceivability argument attempted to show. So, some panpsychist might jump off the boat when they have also to accept this view of causation. Furthermore, panspychists need something much stronger than the claim that causation is categorically grounded on the intrinsic properties of our fundamental entities, they have to hold that causal relations can only have the intrinsic properties of the actual world as their categorical basis. In other words, worlds that verify P are not metaphysically possible. I think that a world that is exactly like ours (consciousness aside) but where fundamental entities lack mentality is conceivable and in this case it is not clear to me how to block the metaphysical possibility of such a world and, importantly, in a way that does not open a possible reply for the materialist –recall that we want the conceivability argument to back up panpsychism. Finally, in the third section Hedda maintains that silicon brains might not be functionally isomorphic to organic brains, if we accept the Russellian conception of the physical. I need some help here, because I do not get the point. Surely biological neurons and silicon chips have very different dispositions and silicon and biological brains have different functions (at more fine grained level), but OI holds that only a subset of these dispostions matters for consciousness, the ones that biological and silicon neurons, ex-hypothese, share. Now I come back to Dave’s 4 notes in post #1, thanks again 1. But the reasons you provide in this page, as I note in the paper, are related to the case of change blindness, where the subject is not attending to the changing feature. To avoid this kind of reply in my example the subject is asked to focus and attend her pain. Do you think that the horrible pain and the no pain at all dancing before the subject “eyes” with her concentrating in her painful experience but not noticing any change is plausible? What do you think about a fading qualia case in the semizombie world? I chose a dancing qualia case with pain and no pain because this is the most pressing case for me. 2. OI* is a weakened version of OI and is the one that rescued the spirit of OI, this requires the experiences to be supersimilar (namely, to respect phenomenal structure). In the version you propose only being conscious is organizational invariant, I did not consider this option because it is too far away from OI. 3. The position that I understand that you are proposing is the one that I consider in p.8 in which RED34 and RED 34* are the same kind of experience. And all that I say is that “this alternative seems to make PP and OI (or OI*) compatible at the price of accepting that fundamental entities do not play any role in determining the kind of experience” 4. Sure, but my reasoning do not requires dancing qualia cases to be impossible (in any sense). The point that I make is that if one is moved by the dancing qualia argument to endorse OI then one should by this very same reason reject panpsychism. If one think that dancing or fading qualia arguments support the claim that OI is true of the actual world, because it is implausible (or because one thinks that they are impossible or even just because they are weird) that such cases occur in the actual world, then one should reject panpsychism for, it is committed to the acceptance of such a cases in the zombie world, one that verifies all the physical truths of the actual world (thereby being equally implausible or weird). A minor point: if by nomologically possible you mean possible according to the laws of Physics then the semi-zombie world is, I think, nomologically possible (the laws of Physics are silent about the intrinsic nature of the fundamental entities) (I have just downloaded Philips comments, I hope to have a rejoin by Monday) 4 rewiten (Apologize. I have submitted it before re-reading). Sure, but my reasoning do not requires dancing qualia cases to be impossible (in any sense). The point that I make is that, if one is moved by the dancing qualia argument to endorse OI, then one should, by this very same reason, reject panpsychism. If one thinks that dancing (or fading) qualia arguments support the claim that OI is true of the actual world because it is implausible (or because one thinks that it is impossible or even just because it would be weird) that such cases occur in the actual world, then one should reject panpsychism. The reason is that panpsychism is committed to the acceptance of such a cases in the semi-zombie world, one that verifies all the physical truths of the actual world (thereby being equally implausible or weird). A minor point: if by nomologically possible you mean possible according to the laws of Physics then the semi-zombie world is, I think, nomologically possible (the laws of Physics are silent about the intrinsic nature of the fundamental entities) Thanks for your comments Philip. Here are some thoughts on them. In reply to the second argument, Philip claims that it rests on a confusion between plausibility and possibility. I think that there is no such a confusion. Philip is completely right in noting that implausibility is all that the argument for OI requires. However, as I replied to Dave in the previous post, my argument merely holds that if one endorses OI because dancing qualia are implausible in the actual world then, for the very same reason, one should reject panspyschism. Dancing and fading qualia are, pre-theoretically, equally plausible or implausible independently on whether the actual world is such that it contains strings and strings- or merely strings. In reply to the first pass, Philip doubts that accepting that the fundamental entities of the actual world only determine the subjective character is a cost for panspychism. I presented it quickly as a cost because in reply to the conceivability argument, panspychism is presented as the claim that the fundamental entities of the actual world determine the phenomenal character of experience. Now, (if I am right) as a consequence of endorsing OI, it results that they can only determine the subjective character but not the qualitative one. Let me try something more because I am now wondering whether this kind of panspychism (call it ‘SCPP’ for subjective constitutive panpsychism and contrast it to CPP that maintains that both subjective and qualitative character are determined by the intrinsic nature of fundamental entities) can block the conceivability argument. Let Q be ‘Philip undergoes a red experience at t’ P and not Q is conceivable and so epistemically possible. CPP can blocks the metaphysical possibility of a world that satisfies P and not Q due to the connection between the intrinsic nature of the fundamental entities of our world and the phenomenal character of experience. But SCPP cannot, I think. The reason is that every world that satisfies P is a world where Philip has an experience, but not every world that satisfies P is a world that satisfies Q –for Philip might have a different experience in worlds that satisfy P. This would defeat materialism as the claim that a minimal duplicate of the actual world is a duplicate simpliciter. If this is right, SCPP would have to give up materialism and here is a clear extra cost. 1. as i say in the book, i think cases of an attended unnoticed change would be stranger than that of an unattended unnoticed change but perhaps not so strange as to be ruled out in all circumstances. i also say that russellian monism provides a natural model in which such changes could occur. 2. your definition of “supersimilar” in the article is much stronger than just “respect phenomenal structure”. i don’t think my response requires supersimilarity in the sense defined in the article, but it might go with supersimilarity in this weaker sense (though i don’t think this weaker sense deserves to be called “supersimilarity”, just something weaker like “isomorphism”). 3. as before, i don’t think your argument against this option or against the full POI option really works because it doesn’t take into account the possibility of a multiple realizablity view. on such a view, the microphenomenal properties will determine the macrophenomenal property, but in a multiply realizable way so that the macrophenomenal properties don’t determine the microphenomenal property. 4. here i agree with philip (and my earlier self in the 1996 book) that the acceptance of fading/dacing qualia cases in the actual world is a much worse consequence than acceptance of fading/dancing qualia cases in a merely possible world such as semi-zombie words. Hi everyone, really great discussion going on in here! Since I am neither a panpsychist or an organizational invariantist I don’t really have a horse in this race but I am very interested in the differences between the dancing and fading qualia arguments. I have always thought that change blindness gave us a reason to suspect the dancing qualia arguments. What I am wondering is why we don’t have similar reasons to suspect the fading qualia argument? Dave considers cases where subjects deny blindness but suggests that they are not fully rational. But don’t we have evidence that fully rational agents make mistakes about their own conscious experience all the time? Here I am thinking, in part, about the kinds of evidence that we get from Eric Schwitzgebel to the effect that naive introspection is not reliable. I would add to this the kinds of Dental Fear cases that David Rosenthal has talked about, and a host of others. Doesn’t this give us reason to suspect that the fading qualia argument doesn’t work? More generally, if we think that empirical discoveries can lead us to be suspicious of one of the arguments why shouldn’t that also undermine our confidence in the other? Besides this I am not really clear why the change blindness considerations don’t by themselves motivate a suspicion of the fading qualia argument. I mean isn’t going from a phenomenal experience as of a loud noise to a softer one a change? And aren’t we comfortable with the idea that a change can go unnoticed (even attended changes can be unnoticed if the conditions are set up right). Maybe the idea is that we are not *completely* wrong in such cases? But if so I don’t see why that is all that more strange than the attended dancing pain cases. That is, why can’t we say this kind of global mistake is strange but not so strange so as to be ruled out in all cases? Moving on to issues that are more directly relevant to Miguel’s argument, I am curious about Dave’s #3, the multiple realizability gambit. How is this supposed to work? I thought the microphenomenal properties were supposed to be such things as red, green, etc. If so is the idea that a macrophenomenal experience of red could be constituted by, say, microphenomenal red, or microphenomeal pink+black, or microphenomenal magenta+yellow? Or is the proposal something else? Finally, regarding 4, I read Miguel as trying to suggest that the semi-zombie world may in fact be the actual world rather than a merely possible one. As he says we don’t have any reason to rule it out do we? This is especially a problem for you if you do not find the dancing qualia argument a compelling reason to think it can’t actually happen. Can you (Dave) tell us why you think the semi-zombie world is merely possible rather than actual? How do we rule out that we are in such a world? Actually, just after I posted this it occurred to me that the experimental results from partial report paradigms should be added to the list above. If we can have generic conscious experience (or degraded conscious experience) that we mistake for specific conscious experience (or non-degraded conscious experience) then shouldn’t that give us reason to suspect the fading qualia argument? I think we all have to accept that there is such phenomenology and that we make these kind of mistakes, especially in the wake of Sid Kouider’s work. Hi everyone, First, thanks Miguel, for your replies to my comments! Here are some answers to your concerns: “I think that Hedda is right that giving up constitution blocks my first objection but it also loses the motivation from the conceivability argument.” If you give up constitution you still have a reply to the conceivability argument. Completely non-experiential zombies are ruled out for the same reasons as before, they verify but don’t satisfy P. Philip’s panpsychist zombies will be metaphysically possible, but nomologically impossible, but that’s ok, because I think there are arguments that a merely nomological connection between macro-consciousness and micro-consciousness would much less problematic for panpsychism than a merely nomological connection between physicSal properties and consciousness would be for physicalism. — “it might help if you could please present an answer along these lines to the dancing qualia case that I present.” I was intending to address your dancing qualia case. Panpsychist should deny that dancing qualia would be the result of this configuration, and say that what would actually happen if one could replace the pain correlate with a strings- realised pain correlate, would be that not only the pain, but also the whole experiential field would disappear. Panpsychists are not committed to saying that dancing qualia is the result of the scenario you describe, so I think they should say that suddenly disappearing qualia would be the result, and this would not be as implausible as in Dave’s original case for the reasons I gave. — “I do not see why such interaction would be problematic. If our Physics go in the right direction P is independent of the intrinsic nature of the fundamental entities.” I don’t think physics entails anything one way or the other here. P (the sentence of all physical truths) is also silent about the nature of causation. I’m arguing that (non-Humean) causation between fundamentally different substances is metaphysically impossible, even though it might still be nomologically possible, insofar as being compatible with P is all that’s required for the latter (got another comment below on that). — “Hedda suggests the possibility that “causal relations are entered into in virtue of the intrinsic properties of the relata, or that dispositions are categorically grounded.” This is already a step far beyond the claim that consciousness depends on the intrinsic nature of the fundamental entities of the actual world that is what the conceivability argument attempted to show.” I agree it doesn’t follow from the conceivability argument alone, but I do think that Russellian monism as a whole entails that there are non-Humean elements to causation. It would be odd to adopt just the idea that “consciousness depends on the intrinsic nature of the fundamental entities of the actual world” and not the other aspects of Russellian monism. — “Finally, in the third section Hedda maintains that silicon brains might not be functionally isomorphic to organic brains, if we accept the Russellian conception of the physical. I need some help here, because I do not get the point.” My point here was just that the fact that the physical is functional all the way down might seem to enable Russellian panspychists to say that OI is trivially true. But the intuition behind OI is that behaviorial functions are what matters, so it doesn’t help much after all. So I didn’t mean to disagree with you, just make clear that it wouldn’t help that much, it seems prima facie that it’s something Russellian panpsychists could make use of. — I also had some question about what Dave brought up. As Miguel and Richard said, it also seems to me that the second pass scenario is nomologically possible, because it’s compatible with the laws of physics of our world, since they are silent about the intrinsic natures of things. Are there other criteria for nomological possibility? I thought maybe one could argue that the scenario was nomologically impossible (not “just” metaphysically impossible as I think it is anyway) for example like this: if we as panpsychists assume that our world is in fact made of just panpsychic experiential stuff (the initial conditions of the universe was a singularity of experiential stuff?), and there might be a kind of conservation law that physics doesn’t describe but is still a law of nature: that nothing mental can be produced by anything non-mental (expresses the anti-emergentism which part of the motivation for panpsychism in the first place), and vice versa, nothing non-mental can produce anything mental. This would rule out the existence of non-mental strings- in our world. I was also, like Richard, curious about how multiple realization could work with micro-/proto-phenomenal properties. It’s clear how abstract/relational properties can be multiply realized by concrete ones, but it’s not as clear to me how concrete/intrinsic properties can be multiply realized by other concrete/intrinsic properties. I wonder whether color mixing examples are proper examples of multiple realization. Either the different color combinations (red+red, pink+black, magenta+yellow) blend and produce the same color (red) without constituting it. Let’s say pink and black makes red, but when we zoom in at the red there are no pink or black grains, just red all the way down. The the pink and black has really disappeared and can’t be regarded as constituents or realizers. On the other hand, if there are such grains, then either there isn’t really any red there being constituted, or all the reds (pink+black, magenta+yellow, etc.) are different when we zoom in so there isn’t really multiple realisation after all. Dave, 1. You do not like the dancing qualia argument anymore, but you do like the fading qualia argument. In this case, as I mentioned, just use the fading qualia argument. It amounts, I think, to the same result. One problem that remains is that I do not understand why change blindness gives any support for a change in mind about the case. Change blindness might show that there are features of experience that we do not notice, but not features that we cannot notice (they still might be available for rational control of action and speech –just attend them and you’ll notice them). Dancing qualia requires that there might be features of experience that cannot be noticed –features of experience that are not accessible in any (relevant) sense. This is by no mean suggested by change blindness cases and for some this idea makes no sense. 2. I was thinking, maybe wrongly, that supersimilarity follows from isomorphy. If Red34 can be replaced by red34* and the phenomenal structure is respected (the space containing red34 and the space containing red 34* are isomorphs), then there is no way to phenomenally distinguishing the two experiences (and therefore they are supersimilar) 3. If microphenomenal properties are just responsible for the subjective character and organization is responsible for the qualitative character (as you first suggested), then we cannot block the conclusion that materialism is false if my reasoning in #5 is right. In your last post you seem to suggest something stronger than what I considered. If this is the case could you please elaborate? 4. I fail to see what justifies your reply. Especially when the actual world might perfectly be a semi-zombie world as Richard also remarks. Let me try in a different way. (Considering that the modality involved is nomological): 1. S believes that dancing and fading qualia scenarios are implausible of the actual world. 2. If one believes that dancing and fading qualia scenarios are implausible of the actual world, then one should not believe that the actual world might be such that dancing and fading qualia obtain in the actual world (OI attempts to prevent these cases) 3. If one believes that panpsychism is true of the actual world then one should believe that the actual world might be such that dancing and fading qualia obtain in the actual world (the actual world might be a semi-zombie world) 4. S should not believe that panpsychism is true of the actual world. Thanks for your comments Richard. One methodological question. There seems to be an important asymmetry (well, I am precisely asking about how important this asymmetry might be) between the cases you present in which, allegedly, the subject is wrong about her own experience and the fading qualia. The reason is that in the fading qualia the subject, after the replacement, cannot be right (is systematically wrong) about her experience and one might find this really odd. What do you think? P.S. Just to solve one doubt that came to my mind reading your comment (and to try to invite a horse of you to the race). If the subject thinks (non-inferentially, etc) that she is having a specific conscious experience then, isn’t it the case that he is having an specific conscious experience? (the content of the HOT determines the phenomenology). He might be wrong about his first order state but not about the phenomenology, which is given by the HOT. Maybe the reply is that the thought I am considering is an inferential one, but I am not sure what would justify this claim. Be that as it may, I am interested in your thoughts (here or somewhere else) about this. P.S.’ As I understand them microphenomenal properties cannot be red or green but properties that when condition C obtains (when properly organized for example) give rise to red or green. Hedda, “If you give up constitution you still have a reply to the conceivability argument. Completely non-experiential zombies are ruled out for the same reasons as before, they verify but don’t satisfy P. Philip’s panpsychist zombies will be metaphysically possible, but nomologically impossible” Materlialism as I presented in the paper is a thesis about metaphysically possible worlds. So, if micro-consciousness is not constitutive of macro-consciousness then there are worlds that satisfy P (and thereby they are constituted by the same fundamental entities that our world –microconsciousness) but do not satisfy Q (Q tells us about macro-consciousness and micro-consciousnes is not constitutive of macro-consciousness) and therefore materialism is false. “But that’s ok, because I think there are arguments that a merely nomological connection between macro-consciousness and micro-consciousness would much less problematic for panpsychism than a merely nomological connection between physical properties and consciousness would be for physicalism.” Sorry, I do not see why. The alleged problem for physicalism is the explanatory gap due the lack of a priori entailment between phenomenal truths and physical truths. The kind of panpsychism you favor has the very same problems between phenomenal truths and micro-consciousness truths. – “Panpsychist should deny that dancing qualia would be the result of this configuration, and say that what would actually happen if one could replace the pain correlate with a strings- realised pain correlate, would be that not only the pain, but also the whole experiential field would disappear.” I think I am missing something. In the example I present, only the core neural correlate of the pain is changed. Why should the whole experiential field disappear? Be that as it may, your proposal might be an interesting way for panpsychism to resist the arguments in favor of OI. In the paper I only attempt to show the incompatibility of OI and PP. My argument is based on the the acceptance of the principles that back up OI. In the case of suddenly dissapearing qualia one has to accept that it is implausible that “the replacement of a single neuron (leaving everything else constant) could be responsible for the vanishing of an entire field of conscious experience. This seems extremely implausible, if not entirely bizarre. If this were possible, there would be brute discontinuities in the laws of nature [which hold in the semi-zombie world] unlike those we find anywhere else.” – “I don’t think physics entails anything one way or the other here. P (the sentence of all physical truths) is also silent about the nature of causation. I’m arguing that (non-Humean) causation between fundamentally different substances is metaphysically impossible, even though it might still be nomologically possible, insofar as being compatible with P is all that’s required for the latter (got another comment below on that).” Nomological possibility entails metaphysical possibility. – “I agree it doesn’t follow from the conceivability argument alone, but I do think that Russellian monism as a whole entails that there are non-Humean elements to causation. It would be odd to adopt just the idea that “consciousness depends on the intrinsic nature of the fundamental entities of the actual world” and not the other aspects of Russellian monism.” Regarding causation I guess we both agree that this form of panpsychism is not required to reply the conceivability argument. Some might disagree about how acceptable this form that emerges is. – “My point here was just that the fact that the physical is functional all the way down might seem to enable Russellian panspychists to say that OI is trivially true. But the intuition behind OI is that behaviorial functions are what matters, so it doesn’t help much after all. So I didn’t mean to disagree with you, just make clear that it wouldn’t help that much, it seems prima facie that it’s something Russellian panpsychists could make use of.” I think we do disagree, for OI is in any case a substantive thesis. But I am still confused. We do not need any principled distinction between realizers and structures. Structures can be described at very different levels. OI selects one level, L, as the one that, at least in the actual world, has to be satisfied for (macro-)consciousness. Whether inferior levels of description are satisfied by an structure is irrelevant. In other words, two structures (in the actual world) are of the same kind with respect to (macro)consciousness iff they satisfy level of description L. Biological brains and Silicon brains are structure of the same kind in this respect. Russellian Monism does not deny that there are different levels of description of structures. – “Are there other criteria for nomological possibility? I thought maybe one could argue that the scenario was nomologically impossible (not “just” metaphysically impossible as I think it is anyway) for example like this: if we as panpsychists assume that our world is in fact made of just panpsychic experiential stuff (the initial conditions of the universe was a singularity of experiential stuff?), and there might be a kind of conservation law that physics doesn’t describe but is still a law of nature: that nothing mental can be produced by anything non-mental (expresses the anti-emergentism which part of the motivation for panpsychism in the first place), and vice versa, nothing non-mental can produce anything mental. This would rule out the existence of non-mental strings- in our world.” It seems to me that the assumption that the world is made just of panpsych experiential stuff is question begging. And the assumption that nothing mental can be produced by anything non-mental ad-hoc (or resting in the constitutive connection between micro-consciousness and causation we have discussed –such that every world that verifies P is a world that satisfies P) In any case, the intuitions about the fading and dancing qualia seem to be pre-theoretical and so there is no reason to change our view on what is nomologically possible. What about the following? Epistemic Principle (EP): If evidence E is compatible with a relevant scenario in which not p, then E does not offer support in favor of p. Consider two scenarios: 1. The fundamental entities of the actual world are strings. 2. The fundamental entities of the actual world are strings and strings-. Let E be whatever evidence we have in the actual world, which, according to the dancing and fading qualia arguments, supports the believe that dancing and fading qualia do not obtain in the world we inhabit. Let p be the proposition that dancing and fading qualia do not obtain in the world we inhabit. E is compatible with 2 and 2 is a relevant scenario. But, if panpsychism is true then, in 2, p is false. Therefore by EP, E does not support p or panpsychism is false. i’ll reply simultaneously to hedda, miguel, and richard. on fading and dancing qualia: the dancing qualia case requires momentary (if large) errors, but there’s a good explanation of those errors in terms of the massive switching process that takes place just at that moment. (n.b. the motivation from change blindness is just a motivation, nothing like a conclusive argument.) the fading qualia case requires huge ongoing errors (a subject believing their consciousness is just like mine when instead it contains just a few bits), and furthermore there’s no analogous explanation of those errors, in that the system will be functionally innocuous throughout. (eric’s cases involve much smaller errors and also plausibly involve functional irrationality.) so i think the implausibility of fading qualia is somewhat greater than that of dancing qualia. that said, i think both arguments are plausibility arguments and not entirely conclusive. for present purposes the more important point is that while there’s prima facie pressure for a panpsychist to reject the dancing qualia argument there’s much less prima facie pressure for them to reject the fading qualia argument. on multiple realizability: richard, i’m thinking of microphenomenal properties not as (phenomenal) red or green much more primitive properties that might combine to yield (phenomenal) red or green and other familiar phenomenal properties. hedda, i think concrete/intrinsic properties can be multiply realized. there are lots of ways to build a sphere, or to build a system with a certain mass, from microconstituents, and in that sense the property of being a sphere or of having a certain mass is multiply realized. likewise i don’t see why a given macrophenomenal property couldn’t be built up in lots of ways from microphenomenal properties (though i agree it’s not obvious that color mixing provides such a case). miguel: regarding your argument, i think on the view in question every world that satisfies P will satisfy Q (it will have the intrinsic microconstituents that yield the subjective character and the organization that yields the qualitative character). on semi-zombies worlds: as philip notes, the fading/dancing qualia arguments (like mahy thought experiments in science) are intended as implausibility arguments rather than impossibility arguments. if they work, their conclusion is that theories entailing FQ and DQ aren’t true — not that they’re not possibly true. so if a certain string/string- theory entails FQ and DQ, and the arguments work, the conclusion is that that theory isn’t true (although it may be possibly true). i don’t really see the force of the claim that it “might” be true. if the thought is that it was antecedently epistemically possible that the theory be true, then fine, but then the FQ/DQ arguments are intended as arguments for why we should reject that epistemic possibility as a claim about the actual world. for related reasons i think one can reject it as a nomological possibility — and n.b. i certainly don’t identify nomological possibility with physical possibility! e.g. on my view zombies are physically possible but nomologically impossible. p.s. regarding miguel’s #13: i think the principle here is implausible. it will rule out standard inductive or abductive reasoning, for example. e.g. seeing a thousand sunrises is supposed to give evidence that the sun will rise on the 1001st day, and therefore evidence against the hypothesis that the sun will rise 1000 times and then not on the 1001st day, even though that hypothesis is consistent with the evidence. Re: Multiple Realizability: To make general sense of the Multiple Realizability proposal here we should probably draw a distinction between Structural Composition and Realization. Carbon is a structural component of Methane. Methane realizes Organic Compound. With this distinction in place, there seem to be two ways for the (constitutivist) panpsychist to make room for something in the vicinity of multiple realizability: either in Structural Composition, or in Realization. I will consider these possibilities in turn and argue that there are difficulties either way in appealing to multiple realization to diffuse the tension Miguel points to. 1) Multiple Realizability in Structural Composition If we are constitutivists, it will be difficult to maintain that different sets A and B of microphenomenal component properties structurally compose to yield the same exact macrophenomenal property. It may be that the same set of properties generates distinct structural composites (cf butane and isobutane) but presumably if two structural composites have distinct components then they are distinct. I can think of two ways we might find something like multiple realizability here nevertheless. First, we might say that Structural Compositions out of microphenomenal components do not yield macrophenomenal properties per se, instead they yield properties that realize macrophenomenal properties, though not themselves phenomenal. Second, we might say that the principles of structural composition wash out certain distinctions among composites, so that there literally are not two distinct possible exact structural composite properties here (one the relevant structural composite from set A, the other, the one from set B) – instead, there is just one, equally well composed out of one set of materials as the other. (By analogy, suppose there turn out to be two types of Carbon atoms, from a string theoretic point of view. The first view then would be that there are two exact methane-ish structural composites here, and neither of them counts as a ‘molecular property’ though both realize the property of being Methane. The second view would be that there is only one structural composite property here, Methane, and its instances can involve the one kind of component as well as the other). The problem with the first way is that it is mysterious how a structural composite of phenomenal properties might fail to be a phenomenal property. A panprotopsychist might go in for that sort of thing, but it is not obvious that a panpsychist could. The problem with the second way is that it involves a funny sort of restriction on structural composition – a restriction that may undermine the Constitutivist’s claim to ontological innocence. 2) Genuine Multiple Realizability Alternatively the panpsychist might allow that different microphenomenal components in general structurally compose to yield different exact macrophenomenal properties – but that these exact macrophenomenal properties often realize the same “higher level” multiply realizable macrophenomenal properties – the way that Methane realizes Being an Organic Compound, but other carbon-involving molecules also do. This is what I would say if I were a panpsychist. But how might this move help the panpsychist reply to Miguel? The panpsychist could say that our ordinary phenomenal concepts pick out the multiply realizable macrophenomenal properties – and that various pairs of exact macrophenomenal properties are not even discriminable (though they are distinct phenomenal properties). This might be the beginning of a reply to Miguel (a way of dissolving the initial tension), if the panpsychist could give us a story about what makes the indiscriminability of some of these distinct, exact macrophenomenal states unproblematic (in the way that dancing qualia cases are allegedly problematic). This, I suspect, is the hard part. If the exact macrophenomenal states that get toggled in dancing qualia cases are to be indiscriminable in the way the panpsychist will already have made allowance for, a whole lot of exact macrophenomenal states with significantly distinct microphenomenal components are going to have to be indiscriminable in this way, and that is going to put a lot of pressure on whatever the story is about why this sort of indiscriminability isnt problematic already. I think this is the real upshot for the (constitutivist) panpsychist (who likes the Dancing Qualia argument). Wow, a lot has happened since last I was here! Related (I think) to Jon’s worry above, I wonder what kind of *phenomenal* properties we are talking about at the microlevel if they are not like phenomenal red, etc. It sound a bit like they might be protophenomenal properties, in which case Miguel’s argument goes through for panpsychism. In regards to the the fading/dancing qualia arguments Dave says (and Miguel suggested this too) “The fading qualia case requires huge ongoing errors (a subject believing their consciousness is just like mine when instead it contains just a few bits), and furthermore there’s no analogous explanation of those errors, in that the system will be functionally innocuous throughout.” I am not sure why we don’t have an analogous explanation. In the partial report experiments subjects feel like they have consciously seen all of the stimuli when they in fact haven’t. The results also suggest that there must be some degraded ‘generic’ content (even those like Ned have to accept that there is some of this in order to account for the results). So now take two cases, one where someone is in fact consciously experiencing all of the detail in the stimuli, and one where they are not but rather have the generic gist content. The two experiences seem to be the same phenomenologically even though they aren’t! Why doesn’t this provide us with the basis for an analogous explanation? On a related note Miguel asks, “If the subject thinks (non-inferentially, etc) that she is having a specific conscious experience then, isn’t it the case that he is having an specific conscious experience? (the content of the HOT determines the phenomenology).” This isn’t quite right in the partial report cases (or even in the kinds of cases that Hakwan and I talk about in our paper on this stuff). The subject is under the impression that they consciously see all of the stimulus (say all of the letters in the Sperling cases), and according to me they *do* see all of the letters, just not in the relevant detail. But they are under the impression that they do consciously see all of the detail. So yes the content of the higher-order state determines the phenomenology, but if the content is “I see three rows of letters and their specific identities” then I will be under the impression that I am consciously experiencing detail that is not there in my conscious experience. Lastly, I am intrigued by Hedda’s suggestion about giving up constitution and in particular when she says “I think there are arguments that a merely nomological connection between macro-consciousness and micro-consciousness would much less problematic for panpsychism than a merely nomological connection between physicSal properties and consciousness would be for physicalism.” Can you elaborate a bit on this Hedda? What are these arguments you allude to? Great discussion, thanks! Let me start by adding a few remarks on Dave’s points: “I think on the view in question every world that satisfies P will satisfy Q (it will have the intrinsic microconstituents that yield the subjective character and the organization that yields the qualitative character).” S has a red quale in w@. How do you rule out a world w in which P is satisfied but where S has a green quale? Intrinsic microconstituents only yield subjective character and therefore there is room for this possibility. This would be, for example, a world with different laws connecting micro and macro consciousness. Regarding the second argument it should be clear by now (I thought that it was from the very begging) that nothing in the argument rest on the (wrong) assumption that FQ/DQ are impossibility arguments. Now, looking for our mutual understanding about what the arguments show, let me try one more time. We both agree (and everyone should) that the fact that there are worlds where FQ/DQ obtain is not a problem at all. I hold and you resist that it is a problem if they can obtain in worlds that verify P (because those worlds will satisfy whatever grounds and provides all the evidence for the believe that FQ/DQ are implausible of this world) You say that “the FQ/DQ arguments are intended as arguments for why we should reject that epistemic possibility as a claim about the actual world.” and I say that they are, at least, arguments for why an inhabitant of a world that verifies P (again, I am assuming that a world that verifies P satisfies whatever grounds and provides all the evidence for the believe that FQ/DQ are implausible of this world) should reject that epistemic possibility as a claim of this world. Do you agree? The next step says that for all we know (and can know), this might be our world. You reject my epistemic principle because it would rule out inductive reasoning, but the problem of the epistemic principle with induction is not the point here. The point is that, no matter how we characterize them, we want our principles on evidence and justification to state that, if E is good evidence for p in the scenario S, and S’ is identical to S in all the relevant aspects (whichever they might be), then E is good evidence for p in S’ too. Now, given that both verify P, the only-string-world and the semi-zombie world are identical in all the relevant epistemic aspects (and both of them are equally good candidates to be actual). Hence, I take it that we have two different alternatives here. We either reject that inhabitants of the only-string-world have good reasons for denying that there are DQ/FQ in the only-string-world, or we accept that semi-zombies have also good reasons for denying that there are DQ/FQ in the semi-zombie world. Hi Jon, thanks for joining the discussion! I find your analysis really interesting and useful. Two questions come to my mind. The first one is whether the multiple realizability in the structural composition case can block the conceivability argument. I don’t think so, and the challenges you remark precisely stress the lack of required connection between micro and macro consciousness. What do you think? The second one relates to Genuine multiple realizability. As I am understanding the case, you are stressing that this position would be committed to what I have called supersimilar experiences (distinct phenomenal properties that we cannot phenomenally distinguish). I think that this commitment is problematic, but you add an additional pressure by comparing it to the DQ/FQ case. I think that panpsychists can offer a reply to the question of what makes them indiscriminable: they are indiscriminable because the qualitative spaces they give rise to are isomorphs. OI* grants that. But there is a legitimate question that remains, one that I think captures the core of your worry (if I am not misunderstanding you): how can the isomorphy of the qualitative spaces be explained in terms of micro consciousness? I’d like to hear your thoughts and whether I am misunderstanding anything. Thanks for your clarification Richard, it helps but I am not sure I can completely understand the position. Please correct me where I go wrong. In the partial report case, the subject has a HOT with the content “I see three rows of letters and their specific identities”. Given that the HO determines the phenomenology, the subject has a phenomenology as of a three rows of letters and their specific identities (just to be clear about what I mean by ‘having a phenomenology as of X’: a subject has a phenomenology as of red iff, according to HOT, the subject has a (right kind of) HOT with the content “I see red”). But the subject, you maintain, is not really seeing three rows of letters and their specific identities. This might perfectly be right but it seems to me that she is not wrong about her phenomenology, about the kind of experience she has (though, if you are right, she is wrong about the kind of first-order state she is in, but this is irrelevant for phenomenology; i.e the kind of experience she has). So that it might become clearer my misunderstanding: As I understand the empty HO case. The subject has the HOT ‘I see red’, she has an experience as of red and she is wrong about the kind of first-order state she is in (none), but not about the kind of experience she has. She can introspect by means of a Higher-order HOT targeting the previous one. As a result she will say ‘I have an experience as of red’ and she will be right. Where is the detail I am missing? Dave — but sphericity is structural and mass is relational for the Russellian at least, and isn’t it in virtue of this that they can be multiply realized? I don’t think I can imagine phenomenal intrinsic properties being multiply realized without getting the kind of problems that the color case gets. If there is no example of a phenomenal case, doesn’t it get a bit mysterian? Or the constitutive panpsychism loses ontological innocence, as Jon put it. Miguel: “The alleged problem for physicalism is the explanatory gap due the lack of a priori entailment between phenomenal truths and physical truths. The kind of panpsychism you favor has the very same problems between phenomenal truths and micro-consciousness truths.” I think the lack of a priori entailment isn’t a problem in itself, it’s only a problem insofar as it indicates that there is no identity or constitution/realization, and in the case of physicalism it seems brute necessitation or brute emergence is the only alternative, which is very problematic. For panpsychism brute emergence/necessitation isn’t necessarily the only alternative to identity or constitution/realization. There is, e.g., an option of mental-mental emergence, which is arguably not brute, because it’s more intelligible how something mental can be produced by something else that’s mental, than how something mental can be produced by something fundamentally non-mental (i.e. something purely structural/functional). The lack of a priori entailment is thus unproblematic insofar as it doesn’t rule out mental-mental emergence or similar relations, if such relations are more intelligible than brute emergence and don’t face other problems. But that’s a big IF of course. I’ll go a little bit into how it can be defended since Richard asked! Richard — I think there are two main problems with a merely nomological connection between (human type/macro-) consciousness and its base, be it the physicSal or the micro/proto-mental. With both problems there is a way for non-constitutive panpsychists to answer them, which is not available to the physicalist. Problem one is intelligibility: we don’t only want to know *that* there is a nomological connection between (macro) consciousness and its base, but how or why it obtains. Galen Strawson argues that it’s in principle unintelligible how the mental can even emerge from the physical, partly because the physical is purely structural and functional, while the mental isn’t. In general many people think mentality just being produced by the physical is somehow intrinsically mysterious, it’s more mysterious than how something physical can produce something else that’s physical, even though we don’t fully understand that either (as Hume pointed out!). If the base is not physical as ordinarily conceived but also micro-mental as in panpsychism, one doesn’t face this kind of problem of intelligibility, derived from a radical difference between producer and the produced. It’s more intelligible how something mental can result from something else that’s mental, than how the mental can result from the fundamentally non-mental. Problem two is causal integration: (macro)-consciousness becomes a nomological dangler and seeming epiphenomenon if it is caused or emerges, instead of being identical or constituted by its physical or micro-mental base and attaches . Here it’s open to a non-constitutive panpsychist to say that macro-consciousness is a fusion of individual micro-consciousnesses that are its basis, so that in a sense it replaces its micro-conscious predecessors instead co-existing alongside them. In this way there is no causal redundancy (there is no longer any micro-consciousness around to compete with macro-consciousness for causal relevance, because micro-consciousness has disappeared by fusing into macro-consciousness). This kind of option is not available to the emergentist physicalist (or dualist for that matter), they can’t say that emergent consciousness replaces its physical nomological antecedent, that would be absurd. Miguel, some more replies: “In the example I present, only the core neural correlate of the pain is changed. Why should the whole experiential field disappear?” I think the entire macro-consciousness would disappear, because we have specified that organizational structure isn’t sufficient for (macro) consciousness, the organizational structure has to be fully realized by proto/micro-experiential matter as well – that’s how we combine OI with panpsychism in the first place. Once we have admitted that proto/micro-experiential realization is necessary, it follows that if the macro-consciousness enabling function becomes only partially realized by non-mental strings- as the pain correlate, then a necessary condition for macro-consciousness is absent. “In the case of suddenly disappearing qualia one has to accept that it is implausible that “the replacement of a single neuron (leaving everything else constant) could be responsible for the vanishing of an entire field of conscious experience. This seems extremely implausible, if not entirely bizarre.” With panpsychism, suddenly disappearing qualia doesn’t involve as strong a discontinuity because the transition between macro-consciousness and micro/proto-consciousness is less of a leap than the leap between consciousness and no consciousness, as in the original example. I also think it’s less bizarre because we have independent reason to think proto/micro-mental realization is necessary so it won’t be as arbitrary. I already said that in my first comments though, so I assume you disagree, but why? “Nomological possibility entails metaphysical possibility.” Not if the nomological possibility is defined as compatibility with the actual laws of physics. There are things that are arguably metaphysically impossible, that physics doesn’t talk about at all. I think non-Humean inter-substantial interaction is arguably such a thing. Other examples: some people might think and argue that, e.g., diabolical evil or libertarian free will is metaphysically incoherent, even though there is nothing in physics that rules it out. “Regarding causation I guess we both agree that this form of panpsychism is not required to reply the conceivability argument.” Even if assumptions about the nature of causation are not required to reply to the conceivability argument, as long as such assumptions go well with panpsychism, I don’t see why the panpsychist can’t adopt them even directly for the purpose of responding to your kind of challenge? Why should panpsychists be restricted to minimal assumptions needed to respond the conceivability argument? “It seems to me that the assumption that the world is made just of panpsych experiential stuff is question begging.” Yes, maybe. That’s why I think the interaction objection is more relevant. “And the assumption that nothing mental can be produced by anything non-mental ad-hoc” It’s certainly not ad hoc, there is a lot of motivation for it quite independent of the question about OI. Strawson defends it in “Realist monism” (2006), for example. Also see my reply to Richard on this. Interesting paper. I have a general question about how to think of the case for the hypothesis H that there can’t be fading qualia in the world. Suppose we think that it is conceivable that H should be false, and indeed that there is a world in which H is false. Then H is contingent, if true. What then is the source of our justification for believing that H is true? Many say it would be odd if H were false, and it is implausible that ours is a world where it is false. But this doesn’t quite say what the source of the justification is. One option is that we have a posteriori justification for believing H. Thus it would be odd if not-H in the sense that not-H has a low probability on our empirical evidence. But it is somewhat hard to see what the relevant empirical evidence could be (though I can think of things to say here). In any case, those who defend the fading qualia argument don’t marshal empirical evidence. So, another option is that we have a priori (but non-decisive) justification for believing H (say because we have a priori justification for believing in the actual world at least changes in qualia are noticed). We are just rationally entitled to assign a high prior probability to H and a low prior probability to not-H. Since (we’re assuming) H is contingent, this would be a case of the contingent a priori. Now of course some philosophers believe in contingent a priori truths. In his comments Phillip helpfully compares H to assumptions that skeptical hypotheses don’t obtain (assumptions like things are typically as they look, so we’re not brains in vats; the future is like the past). There are worlds where the world doesn’t cooperate and these assumptions are false. But, on pain of lapsing into skepticism, we must somehow have some kind of justification for believing that they are true. One view – one solution to skepticism – is that we have a priori justification for believing that they are true (“warrant for free”). We are rationally entitled to assign them high prior probabilities (White, Wright). As Phillip puts it (though I’m not saying he endorsed this view), there are worlds where these assumptions are false, but it is (a priori) implausible to think ours is one. (Of course, this is not the only view. Another view is dogmatism. On this view, our empirical evidence can give us an *a posteriori* justification for accepting these assumptions, by way of “bootstrapping”.) But it is hard to enough to accept these alleged cases of the contingent a priori. If there are worlds where the relevant assumptions are false, how could we have a priori justification to believing that they are actually true? It is even harder to believe that we have a priori justification for believing the contingent truth that fading qualia can’t happen in the actual world. This would be a kind of objectionable epistemic dangler. Hedda, When I say that it is a problem, I only mean that it is a problem because the target of the paper is someone who endorses panpsychism as a consequence of anti-materialist arguments (they might well be other reasons for endorsing panpsychism but the paper has a restricted scope) “I think the entire macro-consciousness would disappear, because we have specified that organizational structure isn’t sufficient for (macro) consciousness, the organizational structure has to be fully realized by proto/micro-experiential matter as well” Imagine that my experience is a of a red apple with Bach playing in the background. Really plausible assumption: the core neural correlate of my experience as of red is different from that of my auditory experience. I replace the core neural correlate of my experience as of red by a non-micro consciousness duplicate, therefore I do not have an experience as of red. Why should my experience as of Bach playing in the background disappear? There is no “red macroconsciousness” but there is a “musical macroconsciousness” “Why should panpsychists be restricted to minimal assumptions needed to respond the conceivability argument?” The don’t have to. I say I agree with you in this. It’s just a question about the commitments of the theory. Dave’s clarification about the distinction between physical possibility and nomological possibility make the discussion unnecessary. At some point in the discussion I made the mistake of mixing physical possibility when talking about nomological possibility but they are clearly different and is the second one the one Dave appeals to. In the paper I do not appeal to the question on whether a semi-zombie world is nomologically possible or not and this is not required. The argument rest on the FQ/DQ arguments being equally acceptable for inhabitants of the the semi-zombie world and inhabitants of the only-string world, as I have tried to make it clearer. Hi Adam, thanks for your comments and compliment. Unfortunatelly, I do not have an answer to the question of what justifies claims like “FQ doesn’t obtain in the actual world” but I am also interested in this discussion. I think we can find some guide to a reply by considering what Dave says in the paper: “the replacement of a single neuron (leaving everything else constant) could be responsible for the vanishing of an entire field of conscious experience. This seems extremely implausible, if not entirely bizarre. If this were possible, there would be brute discontinuities in the laws of nature unlike those we find anywhere else.” One might think that evidence E “there are no brute discontinuities in the laws of nature [in the actual world]” supports the claim that FQ do not obtain in the actual world. But the argument do not support the claim that FQ do not obtain in any world (FQ is not implausible in all world); because, for example, there are world with brute discontinuities in their laws. Much more, of course, need to be said. But I think that the argument I present is independent of the answer to this question. I only require the truth of the assumption that there is something C that justifies it and that C will obtain in any world (I only need two) that verifies P, so that in all these worlds H is equally justified. What is relevant is that both the single-string world and the semi-zombie world verify P. If what justifies H is empirical evidence, then both the single-string world and the semi-zombie world verify this evidence. If it is a priori justification, both inhabitants of the single-string world and the semi-zombie world would be justified in believing that there are no FQ in their world (they are equally “rationally entitled to assign a high prior probability to H and a low prior probability to not-H”) Hi Miguel, thanks for your thoughts. But I thought only “suddenly disappearing qualia” would require brute discontinuities. “Fading qualia” wouldn’t. At least, Dave’s objection to Fading qualia is different, namely, that it is just implausible. I agree with that. But hard to say much more – as I said in my original comment. I think you are right Adam. I was simply taken the suddenly disappearing qualia as part of the FQ argument and for that reason I mentioned that something could be said. Having only the fading part in mind I agree with you. Hi Miguel, In reply to your questions and just to generally clarify what I was trying to say above: I see 3 different ways the multiple realizability suggestion might go. 1) View 1: Two distinct sets of microphenomenal properties, A and B. The elements of A structurally combine to yield (at least) one exact structural composite property, P1 and the elements of B structurally combine to yield (at least) one exact structural composite property, P2, distinct from P1. Neither P1 nor P2 is a phenomenal property (though both have phenomenal properties as constituents) but they both realize the same phenomenal property, Q. This I suspect is closest to Dave’s original multiple realizability suggestion above: if “molecular organization” determines qualitative macrophenomenal properties, then it will be the determinable shared by both P1 and P2 that captures this organization (together with the constraint of having some microphenomenal constituents) that is the most specific macrophenomenal property here, Q. [though note that there are other ways of developing the idea where capturing the organization is not what does the work] This is also a way of fleshing out the option I understand you to be considering toward the end of 3.1 – “a phenomenally conscious state is one that satisfies a certain functional[organizational] role and is made out of the kind of entities that are fundamental in the actual world [has microphenomenal constituents]” — though for the discussion of multiple realizability its important to focus on properties as well as states. My main worry with this proposal was that there is a tension for the panpsychist in making sense of the idea that properties like P1 and P2 are not phenomenal properties. This looks more like panprotopsychism than panpsychism to me. There might also be problems here with strong necessities (that is, things that force one to reject the soundness of the conceivability argument). Maybe for example on Dave’s version there will be an issue with determining the grain of organization that counts – there will be many possible differences in organization below that grain, and I dont see how it would be a priori which is the one that determines the cutoff relevant to phenomenology (though perhaps this will be vague, though for reasons I have given elsewhere I do not favour that option). There will also be a question of how the qualitative macrophenomenal properties might be scrutable from the organization and the fact that there are microphenomenal constituents. So there are two challenges here: first a challenge of accounting for the line of demarcation between protophenomenal realizers and phenomenal realizees, and second a challenge of getting to the phenomenal properties’ descriptions qua phenomenal, from their descriptions qua structural composites (some of this, though, may just be the ordinary combination problem) . 2) View 2: two sets of microphenomenal properties, A and B, but only one structural combination property, Q, which is a macrophenomenal property. Q can be thought of as specifying its constituents disjunctively – specifying relations between elements but not whether those elements have to be from A or from B. Here I was thinking that there would be loss of ontological innocence, as well as something odd about countenancing disjunctive properties without countenancing their disjuncts. I suppose there also might be strong necessities in specifying which of the conceptually possible structural composite properties turn out to be genuine structural composite properties. But I am not sure how this interacts with the normal combination problem. 3) View 3: two sets of microphenomenal properties, A and B, structurally compose to yield to distinct exact macrophenomenal properties, Q1 and Q2, but where Q1 and Q2 both realize some more ‘abstract’ macrophenomenal property, Q – and where we cannot behaviorally discriminate between Q1 and Q2, but we can behaviorally discriminate between Q and ~Q. This is in the ballpark of what you call the supersimilarity option. But I want to steer clear of your notion of ‘phenomenal distinguishability’. I am not sure we need it here and not sure what it adds. I’d rather stick to: ‘Q1 and Q2 make distinct contributions to phenomenology’ (which I take to be equivalent with ‘Q1 and Q2 are non-identical phenomenal properties’) and ‘Q1 and Q2 are behaviorally/functionally indiscriminable’. Here, I do not see any new worries about strong necessities (beyond the standard combination problem of getting us from A and B to Q1 and Q2). I also do not see any incoherence here. My sense is that the tension for the panpsychist (who likes the DQ argument) here is not that there is some kind of incoherence, but that there is an obligation to say why this sort of behaviorally indiscriminable phenomenal difference is not problematic, and it is not obvious how to discharge this obligation. One way of putting this challenge is in terms of your OI* – you might say that the Panpsychist owes us an argument that these behaviorally indiscriminable differences respect “phenomenal structure”. I dont know how fair this is : partly because I do not understand just what you take phenomenal structure / phenomenal isomorphism to be, and partly because it seems to me that the challenge is more general — the panpsychist just has to tell us why this sort of behaviorally indiscriminable phenomenal difference is not problematic. I do not see why a demonstration of phenomenal structure isomorphism is the only way to do so. This is where the DQ stuff comes in. It serves as an intuition pump to show us just how many different exact macrophenomenal properties Qi there are, that will be behaviorally indistinguishable. But more importantly, it serves as a way to get the panpsychist to agree to a standard of what makes a behaviorally indistinguishable phenomenal difference Problematic – it is Problematic if it is sufficiently like a behaviorally indistinguishable toggle between attended foveated vivid red and attended foveated vivid green. I am suggesting that you can skip the discussion of quality space isomorphism, and also the discussion of phenomenal indistinguishability, and just say that the challenge on the panpsychist who accepts that panpsychism gives rise to novel examples of behaviorally indiscriminable phenomenal differences, is to say why these are not bad the way paradigmatic DQ cases are bad. Incidentally I dont think its obvious that the panpsychist could defend a quality space isomorphism claim here, in any intuitive sense of isomorphism. First of all because below a certain grain the constituents may be non-isomorphic. Maybe what it is like to be a neuron is far richer than what it is like to be a silicon chip that plays the neuron role. So at best you might only get a comparatively coarse grained isomorphism. Also there is no guarantee that the laws of mental combination are going to be transitive, commutative, linear, continous — or whatever features our intuitive notion of isomorphism may require. But again this may be asking too much – all that I see the panpsychist (who likes DQ reasoning) owes us here is an account of why this sort of behaviorally indiscriminable phenomenal difference is not problematic the way that dancing qualia cases, intuitively understood, appear to be problematic. (But I still think this may be challenge enough) Final point: the fact that there is something we can call ‘multiple realization’ going on does not appear to add much to the dialectic here – we might say that behaviorally indiscriminable phenomenal differences are not problematic when they are differences that preserve some multiply realizable property of a relatively low level of determinability. But this seems either only to relabel the problem (why is it not problematic if it fits that description?), or to shift it to a problem of accounting for how all of these distinct exact macrophenomenal properties Qi come to all realize the same multiply realizable property of a relatively low level of determinability. (maybe some strong necessity troubles here, maybe not). So my discussion here under 3) is really a discussion of your supersimilarity option generally. on fading qualia and semi-zombies: again, i think the situation here is very much analogous to that with inductive and abductive reasoning and other forms of nondemonstrative reasoning. inductive reasoning supports an inductive conclusion, even though there are counterinductive worlds where that reasoning goes wrong. same for abductive reasoning, same for reasoning about other minds (the possibility of zombie worlds where this reasoning goes wrong doesn’t mean we’re unjustified in thinking there are other minds), and so on. the same goes for fading qualia reasoning and semi-zombie worlds: the fact that the reasoning goes wrong in semi-zombie worlds doesn’t mean that it is bad reasoning. that said, i don’t have a great answer to adam’s question about justification. i think that by and large there’s a principle that says we should accept beings’ reports about their conscious experience in the absence of good reasons to believe otherwise. for me that’s sort of analogous to the principle that we should take our perceptual experience at face value in the absence of reasons to believe otherwise. (see the discussion of these things in the postscript to chapter 2 of “the character of consciousness”.) i think that general principle tends to suggest we should hold that beings along the putative fading qualia spectrum are in fact conscious like us. but i don’t have a really great justification for that principle any more than i have a great justification for the principle in the perceptual case. one thing i’m confident of, though: once these cases are actual and some of our compatriots undergo functionally isomorphic partial neural-silicon replacement, almost everyone will take their reports as pretty conclusive reason to think they’re fully conscious, and the hypotheses that they have faded qualia will be treated the way we treat bizarre skeptical hypotheses. Wow, big discussion here… Just in response to Miguel’s reply to my reply: How is your argument about the world in which strings mingle with strings- supposed to work against the OI panpsychist? Why can’t she just say that such a world is possible, but the actual world isn’t like that. The premise of the argument for OI is just that it’s implausible to suppose that the actual world is like that. I might be missing something… I don’t think the argument you make at the end of you reply works. The OI panpsychist can say that P satisfies Q in virtue of the fact that the categorical nature of the parts *in conjunction with* the functional states necessitates Q. In response to Miguel’s question at #20 Miguel says, In the partial report case, the subject has a HOT with the content “I see three rows of letters and their specific identities”. Given that the HO determines the phenomenology, the subject has a phenomenology as of a three rows of letters and their specific identities…But the subject, you maintain, is not really seeing three rows of letters and their specific identities. This might perfectly be right but it seems to me that she is not wrong about her phenomenology, about the kind of experience she has (though, if you are right, she is wrong about the kind of first-order state she is in, but this is irrelevant for phenomenology; i.e the kind of experience she has). First, the higher-order stuff is not really necessary to the view I was suggesting, though it does provide a nice way of making the thesis clear (and of course I am sympathetic to this view, but I just want to be clear that one need not accept it in order to push the line I am trying to push). There is a sense in which the subject is not wrong (they seem to be consciously seeing the letters and their specific identities and this is in fact the way that things seem to them) but in another sense they are in fact wrong (not about what they are seeing (i.e. not about the first-order state) but) about the content of their conscious experience. They take themselves to be consciously experiencing the actual specific identities of the letters, but they are not. The specific identities may be represented by the first-order state, or that state may be missing, but either way their conscious experience is much more sparse than they take it to be. So suppose that I am having the following higher-order state “I am seeing three row of letters and their specific identities which are Q D R / K H D / G A V” and that someone else is having the higher-order thought “I am seeing three rows of letters and their specific identities which are Q # – / # H D / G – V” versus a third who just has “I am seeing three rows of letters and their specific identities which are # # # / # # # / # # #” (the ‘#’ represents the generic content that is not an actual letter whereas the ‘-‘ represents a degraded or partial letter). Then, there is a sense in which all three of these conscious experiences will seem the same to the subjects (they all take themselves to be experiencing the letters AND their specific identities) but this is really only true for me, partially true for the second, and not at all true for the third. It may seem hard to believe that these conscious experiences all seem the same to the subject but this is exactly what the partial report results seem to suggest! Thank you very much for your clarification and further discussion Jon. It is really helpful for me. Although I agree with much of what you say, let me try to say why I still prefer (to maintain) the talk in terms of supersimilar experiences. As you perfectly note, version 3, do not have any commitment to strong necessities. It is, however, committed to the idea that ‘Q1 and Q2 make distinct contributions to phenomenology’ that is to the rejection of OI. But they can still accept DQ/FQ arguments by accepting that Q1 and Q2 do not lead to different behaviours (in any circumstance). You point out that there is an interesting challenge to PP just by noticing that they owe us an explanation of why this behaviorally indiscriminable differences is not problematic. They have a reply ready, one that I owe to Dave before the conference: it might well be case that the phenomenal structure remains the same (that is to say Q2 relates to all other phenomenal states Q1 relates to in the very same way). This leads us to phenomenal indistinguishability; i.e. supersimilar experiences. One would not respect OI but OI*, something close enough to what we wanted to have. Now, as you stress, panspychists still owe us an explanation of this and it is not obvious that they can. I agree with you and I will stress this challenge in future versions of the paper following your recommendation. Now, imagine that they can meet this challenge. I sustain that this is still problematic. One of the main objections to phenomenal disjuntivism is that they postulate supersimilar experiences. Experiences that are different in kind (a veridical perception and an hallucination for example) but are nonetheless phenomenally indistinguishable. Panpsychists who endorse 3 would have to make a similar commitment. Dave, The principle is not uncontroversial but, as far as I know, something along the lines of the epistemic principle I was presenting is widely accepted in epistemology. It rests on the clarification of what counts as a relevant scenario and there is a vast literature on the topic. Defenders of this principle would maintain that, in most discussions, we can assume that the laws of Physics do not have temporal irregularities or something like that. That is enough for allowing the inductive reasoning we do all the time. Scenarios in which there are this kind of irregularities are not relevant in the situations in which inductive reasoning is perfectly legitimate. In the current debate, worlds in which panpsychism is true and that verify P seem completely relevant (maybe some of them are not but I see no reason to rule out the semi-zombie world). All we need in order to reply to the conceivability argument is to assume that panspsychism is true. My point is that no matter how we refine the notion of “relevant scenario” both the single-string world and the semi-zombie world would be relevant. Do you have any suggestion to rule out any of them as relevant? But you seem to take the route of directly reject anything along the lines of the principle. If the later is the case, I could just weaken my point (a mere retorical clarification: not much if I am right and the principle is widely accepted). In so far as anything along the lines of this principle is accepted, then if one accept the FQ/DQ arguments one should reject PP. What do you think? Again that helps me a lot to clarify the view, Richard. I would like to ask you one further and last question to get my mind really clear around the point. I am understanding the claim that ‘these conscious experiences all seem the same to the subject’ as the claim that the three experiences have the same phenomenology –same phenomenal character. According to the version of HOT you are commenting on, what justifies, according to the theory, the claim that the three experience have the same phenomenology if the HOT is different? Shouldn’t HOT predict that they had different phenomenology? If on the other hand, if I were misunderstanding the claim that ‘these conscious experiences all seem the same to the subject’ as the claim that the three experiences have the same phenomenology. What makes the case relevant for the DQ/FQ experiment, where, for the sake of the reductio, we consider the states to have different phenomenology? To be completely sure: qua representationalist theory, the HOT theory you present would maintain that what matters to phenomenology is the content (correctness conditions), not whether such correctness conditions are satisfied (whether the subject is in the first-order state the HOT dictates or not), right? Philip, “Why can’t she just say that such a world is possible, but the actual world isn’t like that. The premise of the argument for OI is just that it’s implausible to suppose that the actual world is like that.” Of course she can, but what grounds the claim that the actual world is the single-string world and not the semi-zombie world? The claim that otherwise, if panpsychism, is true FQ/DQ would obtain? The FQ/DQ arguments show that the inhabitants of a world that verifies whatever bring us to believe that FQ/DQ do not obtain in the actual world should not believe that FQ/DQ obtain in their world. Both the single-string world and the semi-zombie world satisfy this desiderata. My best attempt to make the point clear, though not sufficiently for someone like Dave, is that both inhabitants of the single-string world and the semi-zombie world would have the same kind of evidence and the same physical laws. Why are we, pretheoretically, supposed to believe that our world is the single-string world and not the semi-zombie world? See also the weakened version in the last reply to Dave. This discussion is very helpful to me to figure out the most effective way to make the point, and I am really thankful to all of you for engaging the discussion. You are completely right Philip in claiming that OI panspsychist can say that P satisfies Q in virtue of the fact that the categorical nature of the parts *in conjunction with* the functional states necessitates Q. But I am targeting at someone that takes the conceivability argument as a motivation for endorsing panspsychism. Non panpsychist materialist maintain that just the functional state necessitates Q, thereby rejecting the conceivability argument. This kind of movements are not allowed in the framework of the discussion here. People in the Matrix have the same evidence as us. Like us, I think they ought not to believe they are in the Matrix (even though they are). Similarly, people in the semi-zombie world have the same evidence as us, and ought not to believe they are in the semi-zombie world (even though they are). Another example: People in the world that was created five minutes ago ought to believe there were dinosaurs, even though there aren’t any, as that’s what the evidence suggests. I suppose I’d like to see in the paper comparison to these other sceptical scenarios, as I think the assumption that these scenarios are not actual is similar to the assumption that FQ/DQ scenarios are not actual (but maybe that’s just my idiosyncratic take on the argument…) Hi Miguel, thanks for the follow up! It is important to keep in mind that nothing about what I am saying depends on any kind of higher-order view. Instead, what I am relying on is what I take the partial report results to point to, and -that- certainly is compatible with a higher-order view but it is also compatible with other views. But to answer your question. The claim is that the states have different phenomenology but that the subjects do not notice this (because they are the same in respect of generic content). They all come away with the same beliefs, namely: that they are consciously seeing all of the letters and their specific identities (when they go to report the identities and fail they are very surprised that they did not actually consciously see the identities). This is why it is relevant to the fading qualia cases. In those cases we have two subjects who differ in their phenomenologies (one is detailed and the other is sparse) but who report the same things. This seems like something that is so bizarre that we can rule it out as actually happening, but my thought was that the partial report cases give us an actual cases where this could be true. In response those who like overflow may insist that that the two experiences are really the same (they consciously represent all of the letters and their identities) and that what differs is the access to those conscious representations. But they can’t do this. Importantly even Ned admits that there must be some kind of generic or fragmented (conscious) representations in these cases because that is the only way to explain the actual performance of the subjects. What is at issue for him is whether the entire conscious experience is like that –that is, whether the task performance is completely driven by unconscious representations– but the line I am pushing can accept that the subjects consciously represent enough of the letters to perform on the task (that is the second case above) but *they still take themselves to have seen all of the letters as well as their specific identities* and they are wrong about that. So whether you accept overflow or not you are committed to the claim that the conscious experience of these subjects is sparser than they think it is (this is why it is important that it is not merely a higher-order response) and if change blindness gives us reason to doubt dancing qualia then partial report should give us reason to doubt the fading qualia argument. Terrific Richard and thanks, I perfectly understand now your point. Good one! Your final conditional make sense to me. As I mentioned, I do not think that change blindness tells us anything about DQ. I think, however, that you have a point independently of the truth of the antecedent (change blindness gives us reason to doubt DQ). The reason is that I am not completely sure I can dismiss the relation between partial reports and FQ in the same line I was rejecting the relation between change blindness and DQ (in the example I stipulated that the subject is asked to attend the stimulus). One question. Don’t you think that this result jeopardizes a HOT theory that maintains that the content of the HOT determines phenomenology while linking report and the content of the relevant HOT as I suggested in the previous post? Thanks Miguel! Your questions were very helpful to me in clarifying what I was trying to say (I am now thinking there may be a paper that comes out of this!). I appreciate you following up on this even though it is not strictly related to the topic of your talk. But I don’t think that there is any serious jeopardy for a higher-order view here. Subjects report that they see all of the letters (and they do, just not in respect of all of the detail that is available, much like when you see something through a partially opaque glass…you see the object just not all of its details) and they report that they seem to see all of the individual identities (and even if they do not actually consciously represent the individual identities that is still the way it seems to them). So, these subjects’ reports are accurate as far as they go (they are having the conscious experience that they say they are). If that were all there were to it then we should think that they do in fact consciously experience the specific identities of the letters, but they cannot report them, so we conclude that they were wrong. But in each case the subjects’ reports are a good guide to what they are consciously experiencing. More generally, the kind of relation I see between consciousness and reporting is this: take subjects at face value unless there is very strong reason not to. So even if we did have to reject a report that would not be a big deal…in practice this rarely happens but maybe Anton’s syndrome is such a case. Philip, I was taking for granted that beliefs should aim at true. I thought this is not controversial, and therefore that people in the Matrix should not believe that are not in the Matrix because this is false. One might argue that there are circumstances under which our beliefs should not aim at true (to avoid unnecessary suffering for instance), but even if that were the case, none of these reason seem to be in play here. That said, I do not see why people in the semi-zombie world should believe they are not in the semi-zombie world. In epistemology, among other things, one tries to understand what counts as evidence for certain hypothesis. In many interesting situations, most outside the philosophical discussion on skepticism, the skeptic scenarios are not relevant. One example, I get mad at my friend for not calling me. The evidence I have is that there is no record of any call in my phone. Surely this is not good evidence if we are discussing about skepticism but it is in this case. I do not even consider the possibility of a world being created five minutes ago. In this circumstances I can assume that Physics go in the right direction describing the laws of our world and that they rule out this possibility. If panpsychism is true, then some of the fundamental entities of the actual world enjoy mentality. Panpsychism (the kind of panspychism we are considering –see the discussion with Hedda) is silent on whether there are more than one fundamental entity in the actual world and on whether all fundamental entities in the actual world enjoy mentality; i.e. silent on whether the actual world is a single-string world or a semi-zombie world for example. When one read the DQ/FQ arguments one does not need to have any thoughts about the truth of panspychism. If one thinks that the DQ/FQ arguments are sound, then one should believe that there is some kind of evidence E that support our believe that FQ/DQ do not obtain in the actual world. If we assume that E obtains both in the single-string world and in the semi-zombie world because both verify P (I am not prejudging which of them satisfies P), then, modulo something along the lines of the epistemic principle, either: 1. Evidence E do not support the claim that there is no FQ/DQ in the actual world if panspychism is true. 2. The semi-zombie world scenario is not relevant in this discussion. But the semi-zombie world is clearly relevant within this discussion. The contrast with the scenario in which the world was created just five minutes ago to evaluate whether I am justified in getting mad at my friend seems just obvious to me (but please continue pressing me if it is not to you or if I am missing your point). I don’t think we have evidence that there are no dancing qualia or fading qualia. I just think we are entitled to assume that there aren’t any, in something like the way we’re entitled to assume that we’re not in the Matrix (what’s our evidence that we’re not in the Matrix?). I think we can have reasons to believe things that are false. Suppose the evidence points to Jones murdering his wife, as there’s only a tiny probability that the DNA traces on the knife belonged to someone else. As it happens, against all the odds (a billion to one, let’s say) the murderer had exactly the same DNA as Jones. Still, in finding Jones guilty, the jurors were believing what they had most reason to believe. But suppose I agree that we can’t have reason to believe something false. In that case, I’d say I have reason to believe that there are no fading/dancing qualia (in the same way as I have reason to believe I’m not in the Matrix). But my unfortunate friend in the string-string world who has dancing/fading qualia, doesn’t have reason to believe there are no fading dancing qualia (just as my unfortunate friend in the Matrix doesn’t have reason to believe she’s not in the Matrix). Our epistemic positions are indiscernible, except that the relevant proposition (there are no dancing/fading qualia) is true in my world and false in her world. A pleasure to discuss this issues with you Richard, it also helps me a lot! Thanks for your follow up Philip, I am using evidence in a broad sense. There is something, E, that “entitles” us so to believe there aren’t any FQ/DQ in the actual world. The interesting question in the debate is whether we should be entitled by E to believe that there are no FQ/DQ in the actual world. The argument I present intends to show precisely that if one believes that PP is true, then contrary to what you assert, one is left without reasons to believe that there are no DQ/FQ in the actual world (I have already mentioned the differences with the skeptic scenario. Again, although in the philosophy class about skepticism we might have to concede that our evidence do not support the claim that we do not live in the matrix, it does in many other circumstance in which considering the matrix case is not relevant). It has been an really great and fruitful discussion !! Thanks to Hedda Hassel and Philip Goff for their helpful comments and also to David Chalmers, Richard Brown, Jonathan Simon and Adam Pautz for their thoughtful contribution in this exchange. The paper would benefit enormously from it. Double thanks to Richard for his work making this awesome conference possible! I hope to see you all in future editions of the Consciousness Online Conference!! Comments are closed.