Consciousness Isn’t Substrate-Neutral: From Dancing Qualia & Epiphenomena to Topology & Accelerators

In this video I explain why substrate neutrality is so appealing to the modern educated mind. I zoom in on the Dancing Qualia argument presented by Chalmers which seems to show that if consciousness/qualia requires a specific substrate, then you can build a system where such qualia is epiphenomenal.

In this video I deconstruct this whole line of reasoning from several complementary points of view. In particular, I explain:

1) How substrate-specific hardware accelerators would generate something akin to a mysterious “consciousness discourse” in organisms that have hybrid computational substrates, with the meta-problem of consciousness (partly) explained via the interaction of two very different computational paradigms that struggle to make sense of each other.

2) How the Slicing Problem gives rise to epiphenomenalism for functionalist / computationalist theories of consciousness. This is as big of a problem, from the complete other side, as Dancing Qualia, yet somehow it doesn’t seem to receive much attention. To avoid epiphenomenalism here you require physical substrate properties to correspond to (at least in magnitude) degrees/amounts of qualia.

3) The idea that you can preserve “organizational invariance” by importing the “causal graph” of the system is question-begging. In particular, it assumes that reality breaks down into bit-sized point-like fundamental interactions between zero-dimensional entities. But this is an interpretation of physical facts, which is put into question by precisely things like field theories of physics (e.g. electromagnetism) and at a much deeper level, things like String Theory, where the substrate of reality is topologically non-trivial.

4) I show that beneath a computationalist frame for consciousness there is an implicit conception of frames of reference that are real from specific “points of view”. But as I explain, it is not possible to bootstrap integrated states out of frames of reference or points of view. Ultimately, any non-trivial integration of information that is happening in these ontologies is a projection of your own mind (you’re borrowing the unity of your consciousness to put together pieces of information that define a frame of reference or point of view!).


5) How the mind uses phenomenal binding for information processing can be explained with the lens of self-organizing principles set up in such a way that “following the valence gradient will take you closer to a state that satisfies the constraints of the problem”. Meaning that the very style of problem solving our experience utilizes has an entirely different logic than classical digital algorithms. No wonder it’s so difficult to square our experience with a computationalist frame of reference!

To end, I encourage the listener to enrich his or her conception of computation to include irreducible integrated states as valid inputs, outputs, and intermediate states. This way we put on the same “computational class” things like quantum computers, non-linear optics, soap bubbles, and yes, DMT entity computing systems 🙂 They all use non-trivially integrated bound states as part of their information processing pipeline.

In aggregate, these points explain why the substrate matters for computation in a way that satisfactorily addresses one of the biggest concerns that there is with this view. Namely, Dancing Qualia leading to epiphenomenalism – which gets turned on its head with the Slicing Problem (turns out computational theories were the epiphenomenalist views all along), self-organizing principles for computation, hybrid computing systems, hardware accelerators, field topology, and the insight that “reality as a causal graph is question-begging”. Reality, is, instead, a network of bound states that can interact in topologically non-trivial ways.

Relevant links:

Leave a Reply