Why do you think consciousness should be explained by quantum phenomena at all?
We already have a perfectly fine, worked-out sort of “dualism”: algorithm vs substrate. When a computer runs a quicksort algorithm, the current pivot position certainly “exists” from the program’s point of view, but trying to find an “ontologically basic” physical entity corresponding to it will make you massively confused! In fact, even trying to determine whether such-and-such machine implements such-and-such algorithm, whether two identical computations count as one, etc. will likely lead you into confusion. Even Eliezer fell for that.
My intuition says that a successful path to solving such questions shouldn’t begin with taking them at face value. Rather, you should try to find a perspective that makes the answer follow (yes, mathematically) from simpler assumptions that don’t take the answer for granted in advance. I’m not very good at this game, but I was pretty happy upon realizing how exactly the appearance of probabilities can arise in a purely deterministic world, and how exactly the appearance of the Born rule might arise from a deterministic MWI world with no such rule. Likewise, I hold that the puzzle of consciousness must someday be solvable normally, by a direct mathematical construction that doesn’t appeal to philosophy. If you don’t see a way, this doesn’t mean there is no way. Free will and probability looked just as mysterious, after all.
My main criterion for whether a computational property is objectively present in a physical system, or is a matter of interpretation, is whether it involves semantics. Pure physics only gives you state machines with no semantics. In this case, I think quicksort comes quite close to being definable at the state-machine level. “List” sounds representational, because usually it means “list of items represented by computational tokens”, but if you think of it simply as a set of physical states with an ordering produced by the intrinsic physical dynamics, then “sorting a list” can refer to a meta-dynamics which rearranges that ordering, and “current pivot position” can be an objective and strictly physical property.
The property dualism I’m talking about occurs when basic sensory qualities like color are identified with such computational properties. Either you end up saying “seeing the color is how it feels”—and “feeling” is the extra, dual property—or you say there’s no “feeling” at all—which is denial that consciousness exists. It would be better to be able to assert identity, but then the elements of a conscious experience can’t really be coarse-grained states of neuronal ensembles, etc—that would restore the dualism.
We need an ontology which contains “experiences” and “appearances” (for these things undoubtedly exist), which doesn’t falsify their character, and which puts them in interaction with the atomic aggregates we know as neurons, which presumably also exist. Substance dualism was the classic way to do this—the soul interacting with the pineal gland, as in Descartes. The baroque quantum monadology I’ve hinted at, is the only way I know to introduce consciousness into physical causality that avoids both substance dualism and property dualism. Maybe there’s some other way to do it, but it’s going to be even weirder, and seems like it should still involve what we would now call quantum effects, because the classical ontology just does not contain minds.
I identify with your desire to solve the problem “mathematically” to a certain point. Husserl, the phenomenologist, said that distinct ontological categories are to be understood by different “eidetic sciences”. Mathematics, logic, computer science, theoretical physics, and maybe a few other disciplines like decision theory, probability theory, and neoclassical economics, are all eidetic. Husserl’s proposition was that there should also be eidetic sciences for all the problematic aspects of consciousness. Phenomenology itself was supposed to be the eidetic science of consciousness, as well as the wellspring of the other eidetic sciences, because all ontology derives from phenomenology somehow, and the eidetic sciences study “regional ontologies”, aspects of being.
The idea is not that everything about reality is to be discovered apriori and through introspection. Facts still have to come through experience. But experience takes a variety of forms: along with sensory experience, there’s logical experience, reflective experience, and perhaps others. Of these, reflective experience is the essence of phenomenology, and the key to developing new eidetic sciences; that is, to developing the concepts and methods appropriate to the ontological aspects that remain untheorized, undertheorized, or badly theorized. We need new ideas in at least two areas: the description of consciousness, and the ontology of the conscious object. We need new and better ideas about what sort of a thing could “be conscious”, “have experiences” like the ones we have, and fit into a larger causal matrix. And then we need to rethink physical ontology so that it contains such things. Right now, as I keep asserting, we are stuck with property dualism because the things of physics, in any combination, are fundamentally unlike the thing that is conscious, and so an assertion of identity is not possible.
For more detail, see everything else I’ve written on this site, or wait for the promised paper. :-)
Why do you think consciousness should be explained by quantum phenomena at all?
We already have a perfectly fine, worked-out sort of “dualism”: algorithm vs substrate. When a computer runs a quicksort algorithm, the current pivot position certainly “exists” from the program’s point of view, but trying to find an “ontologically basic” physical entity corresponding to it will make you massively confused! In fact, even trying to determine whether such-and-such machine implements such-and-such algorithm, whether two identical computations count as one, etc. will likely lead you into confusion. Even Eliezer fell for that.
My intuition says that a successful path to solving such questions shouldn’t begin with taking them at face value. Rather, you should try to find a perspective that makes the answer follow (yes, mathematically) from simpler assumptions that don’t take the answer for granted in advance. I’m not very good at this game, but I was pretty happy upon realizing how exactly the appearance of probabilities can arise in a purely deterministic world, and how exactly the appearance of the Born rule might arise from a deterministic MWI world with no such rule. Likewise, I hold that the puzzle of consciousness must someday be solvable normally, by a direct mathematical construction that doesn’t appeal to philosophy. If you don’t see a way, this doesn’t mean there is no way. Free will and probability looked just as mysterious, after all.
My main criterion for whether a computational property is objectively present in a physical system, or is a matter of interpretation, is whether it involves semantics. Pure physics only gives you state machines with no semantics. In this case, I think quicksort comes quite close to being definable at the state-machine level. “List” sounds representational, because usually it means “list of items represented by computational tokens”, but if you think of it simply as a set of physical states with an ordering produced by the intrinsic physical dynamics, then “sorting a list” can refer to a meta-dynamics which rearranges that ordering, and “current pivot position” can be an objective and strictly physical property.
The property dualism I’m talking about occurs when basic sensory qualities like color are identified with such computational properties. Either you end up saying “seeing the color is how it feels”—and “feeling” is the extra, dual property—or you say there’s no “feeling” at all—which is denial that consciousness exists. It would be better to be able to assert identity, but then the elements of a conscious experience can’t really be coarse-grained states of neuronal ensembles, etc—that would restore the dualism.
We need an ontology which contains “experiences” and “appearances” (for these things undoubtedly exist), which doesn’t falsify their character, and which puts them in interaction with the atomic aggregates we know as neurons, which presumably also exist. Substance dualism was the classic way to do this—the soul interacting with the pineal gland, as in Descartes. The baroque quantum monadology I’ve hinted at, is the only way I know to introduce consciousness into physical causality that avoids both substance dualism and property dualism. Maybe there’s some other way to do it, but it’s going to be even weirder, and seems like it should still involve what we would now call quantum effects, because the classical ontology just does not contain minds.
I identify with your desire to solve the problem “mathematically” to a certain point. Husserl, the phenomenologist, said that distinct ontological categories are to be understood by different “eidetic sciences”. Mathematics, logic, computer science, theoretical physics, and maybe a few other disciplines like decision theory, probability theory, and neoclassical economics, are all eidetic. Husserl’s proposition was that there should also be eidetic sciences for all the problematic aspects of consciousness. Phenomenology itself was supposed to be the eidetic science of consciousness, as well as the wellspring of the other eidetic sciences, because all ontology derives from phenomenology somehow, and the eidetic sciences study “regional ontologies”, aspects of being.
The idea is not that everything about reality is to be discovered apriori and through introspection. Facts still have to come through experience. But experience takes a variety of forms: along with sensory experience, there’s logical experience, reflective experience, and perhaps others. Of these, reflective experience is the essence of phenomenology, and the key to developing new eidetic sciences; that is, to developing the concepts and methods appropriate to the ontological aspects that remain untheorized, undertheorized, or badly theorized. We need new ideas in at least two areas: the description of consciousness, and the ontology of the conscious object. We need new and better ideas about what sort of a thing could “be conscious”, “have experiences” like the ones we have, and fit into a larger causal matrix. And then we need to rethink physical ontology so that it contains such things. Right now, as I keep asserting, we are stuck with property dualism because the things of physics, in any combination, are fundamentally unlike the thing that is conscious, and so an assertion of identity is not possible.
For more detail, see everything else I’ve written on this site, or wait for the promised paper. :-)