[Epistemic status: This is a fictional dialogue which surveys the main moves in the literature between Camp #1 and Camp #2 positions described in Rafael Harth’s post Why it’s so hard to talk about consciousness. The goal of this post is to present both sides as engaging in a charitable dialogue and trying to identify the crux.]
C2: I think consciousness is pretty weird.
C1: What do you mean by weird?
C2: Normally, you can describe things in terms of the physical relationships that they have with each other. But when I’m having an experience – if I look at a tomato or an apple – I experience something red, and it feels like that redness is over and above what can be described in terms of physics.
C1: See, I don’t think it’s that weird. I think that our brains have this higher-dimensional representation of different concepts that are all related in a kind of web. Obviously, we don’t have direct access to the high dimensional detail of every neuron firing, so our brain compresses this high-dimensional space into something that’s easier for us to represent to ourselves. When you see something red, this is just the brain picking out a representation for some higher-dimensional concept and compressing it and presenting it to you.
C2: Just so I understand your view. Are you saying that hard problem intuitions are simply confused i.e. that once we understand the full neuroscience, the mystery dissolves? Or are you granting that there’s a genuine explanatory gap but denying it reflects a gap in nature?
C1: The latter. I’ll grant that there’s something to explain. But I think the explanation bottoms out in functional organisation. The hard problem just boils down to a limitation in our introspective access.
C2: I don’t think that fully explains it, though.
C1: Why not? When you talk about the hard problem of consciousness, you’re just saying that it’s difficult to explain from a third-person perspective, but when the brain starts representing things to itself, it generates a first-person perspective – a concept of “self” that things are happening to. This representation is recognised from a first-person perspective inside that self-concept. So the difference between a first-person and a third-person point of view gives rise to the hard problem intuitions because you can always imagine removing yourself from the first person perspective. There’s no actual underlying mystery.
C2: So you’re talking about a phenomenal concept. The brain has a concept of something “red”, represents this concept to itself and this is supposed to generate our experience?
C1: Yes, exactly.
C2: I have a lot of problems with this phenomenal concept strategy. How exactly do you define a phenomenal concept?
C1: You could define them recognitionally – like I recognise a dog, this fits into the causal web of concepts that I’ve built up in my brain and so it picks out a concept that I recognise and deploys it in my brain as a phenomenal concept.
C2: This doesn’t quite work. What about things that you don’t recognise? Are you not conscious of them? For something like colour how do you ever get your first experience of colour if it always needs to be a recognitional concept.
C1: Okay, fair enough. Maybe you could define it quotationally? e.g. redness is [insert red picture here].
C2: Again, wouldn’t you need access to the [insert red picture here] concept in the first place? It seems circular because then redness wouldn’t even be propositional, you’d need to experience an example of it to be able to deploy it.
C1: Alright, I agree that this is a bit cumbersome. But I have a cleaner version which I think you’ll appreciate. You could imagine it as an indexical fact. I could index redness and say “redness is that sort of thing,” where “that” is just the lower-dimensional representation that my brain is presenting to me as a compression of the higher dimensional space.
C2: Okay, I can get on board with this. So I’ve got these indexical phenomenal concepts. Now, I could imagine that a zombie would also possess these indexical phenomenal concepts because a zombie is also in possession of indexicals. It can index and say “this thing, that sort of thing,” etc..
C1: Sure, I agree with that.
C2: But then if a zombie possesses the concepts, then zombies have experiences, which is a contradiction.
C1: Okay fine, let’s say that zombies don’t possess them. Maybe zombies lack the capability to index “redness” but can index other things.
C2: Well, if a zombie doesn’t possess them, then you’re not really explaining anything. If a zombie doesn’t possess this phenomenal concept, then it’s not really powerful enough to solve the hard problem, right? You need something to bridge that gap.
C1: I think there’s an obvious conclusion here. This zombie thought experiment has a problem because you’ve just derived a contradiction from it. Instead of saying that there’s something mysterious to consciousness, why don’t we just say that there’s a problem with this thought experiment regarding zombies?
C2: Fine. I’ll grant that zombies are difficult to conceive. Here’s a weaker form of the thought experiment that I think you’ll be happy to grant. Imagine spectrum inversion. There’s a physical duplicate of yourself, but instead of seeing red where you see red, they actually see blue or green. Your spectra are completely inverted, and you would never know because you’re not able to communicate this difference to each other.
C1: Hold on. You can’t just do a simple remapping between red and blue; this would throw out all sorts of behavioural relations which arise because of the structure of colour space. For example, we have particular light cones in our eyes with particular densities that react to certain wavelengths of light. Surfaces have different properties of reflectance which interact in complex ways with these wavelengths. There is also significantly different wiring in our visual cortices which will produce behavioural differences. I don’t accept that you can simply invert the spectrum here and leave behaviour unchanged.
C2: Good point. There’s obviously a lot of structure which needs to be respected. But you need something stronger for your view to go through. You need the full physics to be fixed, and for the feeling of blue or the feeling of red to also be fixed by the physics.
My claim is the following: I could conceive of the feeling of red or the feeling of blue being slightly different, maybe some rotation in colour space that takes you a little bit away from redness and towards blueness; some difference in the qualitative character of my experience with the exact same physics.
C1: I’m not convinced this actually makes a difference like you claimed. If we permute our spectra or invert them and somehow miraculously leave behaviour unchanged. That’s not creating an actual difference in the world that can be observed.
C2: Hold on, you just admitted that your hard problem intuitions fall out of the difference between third-person and first-person relationships. So it might not make a difference in the third-person observable behavior, but it would make a difference to the first-person experiencer. They would feel something different if their experience were different.
C1: A lot of work is going into the word “difference” there. What are you actually claiming is the difference?
C2: Imagine there’s a person at time T experiencing redness. And then at time T2, the exact same physical structure is replicated, but instead of redness, they are now experiencing blueness while the underlying physical relations remain fully fixed. This is conceivable.
C1: This is perfectly compatible with my position. You’re saying that you can hold this physics constant and vary the experience without changing my behavior? But the experience has now changed its relationship with my memories so my behaviour would change. I would now remember the red experience so if the experience really did change I would say “Oh, I’m experiencing a blue experience now!” which changes the underlying behaviour.
C2: Okay, fine. Let me try something stronger. Imagine there’s world A where the physics is fixed and I’m experiencing a red experience, and world B where the physics is fixed and I’m experiencing a blue experience. World A and world B have the exact same physical structure, but I claim that they are different worlds because my experience differs between them.
C1: I disagree. From my perspective, they are the same world, and I don’t see any substantial actual difference in these two worlds. If you think there is a difference, I’d like to know what exact physical fact makes the difference.
C2: This is a good challenge. Now let me go on a little bit of a tangent. What do you mean by physical fact?
C1: That which is described by our physical laws.
C2: Good. So a physical law would be describing, say, how an electron moves in an electric field?
C1: Yes, exactly.
C2: Great. So what is charge? How do you define charge?
C1: Charge is defined with respect to the equations of electromagnetic interaction. For example it appears in Coulomb’s law.
C2: Good. Now how do you define an electric field?
C1: Electric fields deflect charge. Electric field and charge are related by physical laws and it’s the physical laws which give you this structure.
C2: Isn’t this kind’ve circular? You’re defining charge with respect to electric fields and vice-versa.
C1: It’s not really circular – it’s a web of structural relations. Each entity in the structure is defined by the say it’s related to other entities in the structure.
C2: So if I had an electron and it wasn’t actually in an electric field, is it still charged?
C1: Yes.
C2: But it’s not being deflected at that moment. How can you tell it’s charged?
C1: I would say counterfactually that it would have been deflected if it had been in an electric field even though none was there at the time.
C2: Okay. So you’re essentially saying that the structure can fully specify the underlying reality. All I need to do is write down the equations and the relations that are described by the equations fully underpin reality.
C1: Yes.
C2: In world A and world B, all of these structural relations are identical, so that’s why you want to say that the two worlds are identical.
C1: Yes.
C2: Well, I would say that world A is different precisely because I am having a different experience in world A to world B, and counterfactually, the first-person observer would realise that if they were in world B, they would be having a different experience. My claim is counterfactual: if the observer switched, then they would realise that they are having different experiences. That’s what I mean by saying they are different worlds.
C1: I appreciate what you’re doing, but I’m not really convinced that this is modally potent. You’re trying to cash out this counterfactual in terms of something that is not metaphysically possible to achieve. In physics, things are really happening in the physical world. You’re talking about things which can never happen and are not even metaphysically possible.
C2: Well, I think it is modally potent, and I’m cashing it out in precisely the structural terms that you seem to care about. You were happy to use counterfactuals to define the electrons’ charge even when it’s not in an electric field. Why not now?
C1: Well our counterfactuals are importantly different. I’m cashing it out in terms of what an electron would do in an electric field. Your counterfactual doesn’t specify how the system would do anything differently because both worlds have the exact same laws of physics. There are no causal powers at play and you can’t ever realise this counterfactual because it’s not metaphysically possible to actually switch world A with world B.
C2: It has counterfactual causal powers though. If you switched them then you would see a difference from your first person perspective. You just can’t do the switch in practice.
C1: Why go to all of this work to posit a mysterious metaphysical counterfactual that doesn’t seem to have any real causal powers though? Shouldn’t the more parsimonious move be to reject the need for this counterfactual?
C2: Sure, I’m all for parsimony if we can account for all of the data – but this purely relational view of experience doesn’t explain the first person data points of my experience. Something is needed to fix what blue and red feel like from my first person perspective. If nothing fixes this then there are redundant degrees of freedom.
C1: I’m glad you bring up redundant degrees of freedom because I think that’s exactly what you’re focussing on here. There are many equivalent redundant descriptions of reality that are all correct. It’s like choosing a coordinate system or gauge potential. The actual choice doesn’t affect the physics.
C2: I grant that the choice doesn’t affect physics (third person) but it does affect phenomenal experience (first person). Phenomenal character is fixed by that choice among the redundant degrees of freedom!
C1: I think we’ve gone way off the metaphysical rails here but let me play along for the sake of argument. What exactly is fixing the experience here? It sounds like magic.
C2: Well when you described the electron moving through the electric field you described it in purely relational terms using an equation and counterfactual structure. But you said nothing about what charge really is intrinsically.
C1: What do you mean?
C2: I mean there’s an extra property not captured by the relation alone. If you switched charge with some other property (call it scharge) standard physics would say this doesn’t matter as it doesn’t affect the equations of motion. But on my view it does matter. There’s an intrinsic property which has changed which sits separately to the relational properties described by the equations of physics.
C1: And let me guess… this intrinsic property is non-physical right?
C2: Well it depends what you mean. I’d say it is physical as it’s contained in the physical particles and fields themselves. It’s just not relational like the properties of standard physics.
C1: Why postulate such a property? If it doesn’t have any physical observables it would be causally idle.
C2: Well it’s not causal in a relational sense. Physics is still causally closed. It’s just that the intrinsic properties realise this causal structure.
C1: It sounds epiphenomenal to me. You couldn’t explain any observable behaviour by postulating these properties.
C2: Sure, it wouldn’t have any third person physical observables. But it would do the job of fixing the phenomenal character of experience from a first person perspective.
C1: And every physical particle has these intrinsic properties.
C2: Yes.
C1: So everything is conscious then?
C2: Uhhh…
C1: Well this is the logical conclusion of your view! You’re saying electrons have a categorical property which fixes the phenomenal character of your experience. So electrons are conscious! That sounds absurd.
C2: Sure. But absurd sounding things can be true all the same. Quantum mechanics and General relativity sounded absurd at the time.
C1: Let me put it this way then, how do you combine all of these tiny little microexperiences into a coherent macroexperience? You have a combination problem.
C2: Okay, I don’t know, but at least this gives us a good starting for solving the hard problem right? We’ve reduced the hard problem into a combination problem. That seems more tractable.
C1: You’ll forgive me for not being convinced. If we have to postulate conscious electrons I think we’ve gone off the rails. I can just reject your move to make World A and B counterfactually different even though it’s not metaphysically possible to switch them.
C2: Okay, fine. Instead of saying these categorical properties are phenomenal let’s say they’re proto-phenomenal. They only yield the full suite of phenomenal experience when combined in the appropriate way but observed in isolation they’re not phenomenal.
C1: Oh, come on! We got into this mess in the first place because we weren’t happy bridging the non-phenomenal/phenomenal gap using physics and structure alone. Now you’ve just re-introduced the gap except we now have these causally idle proto-phenomenal properties!
C2: I think it makes sense though! The problem with your view is that there are structural relations all the way down and there’s nothing to realise the structure. On my view we at least have a base which realises the structural relations.
C1: I disagree that a base is needed for structure to be realised. And I think this is the crux between us.
C2: How would you have structure without a realiser though?
C1: Well consider a chess game. There are rules which define the way the pieces move; bishops move diagonally and rooks move in straight lines. The rules of the game completely specify “chess”. So much that you could even remove the pieces and just play blindfolded chess. As long as the rules are intact you don’t need any mysterious base to realise chess.
C2: This is a great analogy. If I switch the rook and bishop pieces on the board but keep the rules the same something has obviously changed.
C1: Nothing has changed about chess qua chess though! It’s still chess!
C2: Yes, because chess is fully specified by structure. Consciousness isn’t fully specified by structure – there’s something else in addition to the structure and that is the categorical base which is realising it.
C1: But to my previous point you can imagine the base being absent and still playing blindfolded chess. The fact that chess is structural doesn’t need a base to realise it.
C2: Well, blindfolded chess still requires players with minds keeping track of the board state. The physical pieces are inessential, but something has to realise the structure. You’ve just shifted the realiser from wooden pieces to mental states. If you removed the players too, there’d be no chess game – just an abstract mathematical structure that could describe a chess game. The game itself is constituted by something concrete.
C1: I’d say the abstract structure is the game. The players are just one way of instantiating it, but the structure is what matters.
C2: Then we have a genuine disagreement about the ontology of structure. I’d say abstract structures don’t exist independently, they’re always structures of something. You apparently think structure can float free.
C1: Structure doesn’t “float free” in a spooky way. The physical relations between neurons are abstract but it’s the structural relations between neurons which are important for consciousness.
C2: Right, but “relations between neurons” presupposes neurons – things standing in relations. And my question is what neurons are, intrinsically, beyond their relational properties.
C1: There’s no intrinsic property of neurons. They’re just related to each other by physical laws, it’s the relations which are fundamental.
C2: I find that incoherent. Relations need relata.
C1: Let me put it this way, do you believe in chairs?
C2: What?
C1: Chairs. Like sofas, couches, stools etc.. do you think they exist?
C2: Sure, but what does this have to do with anything?
C1: Well, is there an intrinsic property of chairness?
C2: I see what you’re getting at. No. The concept of a “chair” is fully specified by the causal/functional role it plays. But consciousness is different.
C1: How exactly is it different?
C2: Because there’s something about my phenomenal character which is fixed. The redness of red, blueness of blue etc..
C1: This seems like special pleading. Everything we encounter in everyday life can be described by the causal/functional role. The rules of chess, but also life, fire etc.. People historically thought these also needed categorical/intrinsic properties to describe like phlogiston or élan vital and we discovered through continued scientific enquiry that they weren’t needed for a complete explanation of the world.
C2: Yes, but I have a principled reason to special plead here. The complete description of the world is only complete from the third person perspective. It’s incomplete from a first person perspective because we need to explain the phenomenal character of consciousness.
C1: Well why don’t things like chairs, life or fire need a categorical property to realise them then?
C2: Because their definition is fully exhausted by the causal/functional role that they play in the structure of the world. So you can get a complete description of life by just understanding more about the functional roles and relational properties. The same isn’t true of consciousness.
C1: I think we’ve reached an impasse, because I think you can do this for consciousness. Consciousness just is the causal/functional role of the physical neurons in the brain. There’s no additional property needed to fix phenomenal character.
C2: I agree that we’ve reached an impasse here, on my view an intrinsic property is needed to fix the phenomenal character. Whereas on your view there’s just structural relations all the way down, nothing is needed to realise the structure as long as the relations hold.
C1: I agree that this is the crux.