Here is David Pearce on the phenomenal binding problem:
Mankind’s most successful story of the world, natural science, leaves the existence of consciousness wholly unexplained. The phenomenal binding problem deepens the mystery. Neither classical nor quantum physics seem to allow the binding of distributively processed neuronal micro-experiences into unitary experiential objects apprehended by a unitary phenomenal self.
I am increasingly of the view that there is no such thing in Experience as unitary experiential objects, for the same reason that I think closed individualism is false.
If we suspect that an Experience E “listens” to whether Binding B had happened, then we call B a cause of E and draw an arrow from B to E in a causal diagram. Naturally, the answer to our query about B and E is likely to depend on other variables as well, which must also be represented in the diagram along with their causes and effects.
The listening pattern prescribed by the paths of the causal model usually results in observable patterns or dependencies in the data. These patterns are called “testable implications” because they can be used for testing the model. These are statements like “There is no path connecting B and E,” which translates to a statistical statement, “B and E are independent,” that is, finding B does not change the likelihood of E. If the data contradict this implication, then we need to revise our model. Such revisions require another engine, which obtains its inputs from testable implication outputs and data inputs and computes the “degree of fitness,” that is, the degree to which the Data are compatible with the model’s assumptions.
The trouble with supporting binding as a causal precursor to experience is that we cannot collect data in the form of crisp instances of “phenomenally-bound objects”/”slices of experience with well-defined boundaries.” Not because introspective evidence doesn’t count, as I very well think it should, but because this is one human-universal case of hemispatial neglect. In other words, I am fairly certain that no one can capture a snapshot of “now” within experience. It may be possible to be a Level 4 Multiverse creature that can see all our frames of experience laid out as a glittery mat, perhaps even with some semblance of contiguity between some frames and not others. However, it seems fair to assume that we are not such beings, and since we are “in the inside,” it is perhaps impossible to ever collect an instance of binding.
First to clarify what we mean by the binding problem since the binding problem is a term used at the interface between neuroscience, cognitive science and philosophy of mind that has multiple meanings: here I take it to mean the sum problem of how yellow triangles, blue circles, objects, background and abstract or emotional features are combined into a single experience. Since, after all, the processing for blue is smeared across different parts of spacetime than circle, and this is the case for all contents of experience. How do these distributed processes “know” to exclude in their final manifestation those computations from too long ago, or from one brain and not another.
Here is where I take issue with the framing. This binding problem assumes that “single experience” is data. Yet in so far as I support empty individualism, it is as a metaphor to motivate a more physically accurate picture of reality which takes into account its necessary timelessness. But we must not take the metaphor of point experiences literally. After all, we could just as arbitrarily have chosen another mathematical construct such as Whitehead’s point-free geometry. And perhaps the binding problem boils down to there being no fact of the matter as to yellow triangle here, blue circle there, and it is instead the case that the isomorphy truly chosen by the universe would explain it. It could also be the case that we should trust phenomenology more, even if it doesn’t fit into an inference engine, and hence construct a different playing field for truth claims which is not subsumed by causal inference a la Pearl. The danger with this is that there would be no principled way to stop at “yellow triangle really exists” and not just skate past into solipsism – I mean, might as well at that point.
From the outside-in, open/empty individualism is implied. We need not deny Experience, only unitary experiential objects, and only in this precise sense can we therefore call individual consciousness an illusion. But if we are phenomenologists, we don’t reason from the outside-in and instead do so from the inside-out. The problem with this approach is that closed individualism may seem to a phenomenologist just as credible as the moment’s “unity of experience” (a case of closed individualism with smaller Δt). And what’s more, if I am not constrained entirely by a third-person scientific model, I see no reason to lampoon solipsism. The most primordial fact seems to be that this exists; not you, not you, not you, this.
Both extremes seem crazy to most people who essentially believe in individual souls on parallel, linear journeys (normie atheists believe in line-segments: the brain-encapsulated soul began at some point and is obliterated at another; everyone else believes in rays or lines: it continues in some way). A physicalist submitting queries on the matter to an inference engine in the form of scientific questions, comes to realize that there is no non-arbitrary way to draw bounds on an experiential self or micro-subjects of experience. One can postulate a soul using causal vocabulary, that is not the issue – that our common-sense intuitions couldn’t be right in principle – no, the issue for the soul hypothesis is that the scientific trek so far hasn’t found support for it. Hence comes the strictly physical, but often counterintuitive, notion that Experience was never born and never dies because that Experience is no more here than there; it is here and there – there are no lines for independent consciousness orbs.
Even if our model of reality showed that purple circle P and red triangle R depend on a third variable binding B, since we do not have any way to measure B, the query P(R|do(P)) cannot be answered. In that case, it is a waste of time to collect data. Instead, we would need to go back and refine the model, either by adding new scientific knowledge that might allow us to estimate B or by making simplifying assumptions (at the risk of being wrong) – for example, that the effect of B on P is non-existent. In other words that binding doesn’t exist.