The way this transmigration argument is used is a fallacy, and simply reveals ignorance of the way a unified world of consciousness works or how we could be consciously aware of things. The "Nature Has Ineffable Phenomenal Properties" camp
describes one possible theory which if true would clearly invalidate this argument.
The bottom line is, this theory predicts that certain neural conscious correlates predictably and reliably always have the same phenomenal property. If you use some other matter, in some other state, though it may be considered to be behaving like the original, it is fundamentally and phenomenally not like the original. Nature is such that these phenomenal neural correlates can be bound together into a unified world of awareness where we are aware of all of them, and their phenomenal differences, together.
Evolution has simply used these fact to implement our unified consciousness knowledge. This is very much different than abstract computer system for which, by design, what they are like doesn't matter. So nothing, other than what is phenomenally like 'red' could be used to present into the conscious world the same phenomenal likeness. And certainly, nothing that doesn't matter what the representations are like, could enable the system to say of any such merely properly behaving correlate - that is the same as real phenomenal red.
So, this theory predicts that once you remove the correlate that represents a spot of red on the strawberry, if whatever you replace it with is not fundamentally and phenomenally the same, and able to present something that is phenomenally like 'red' to the system, transmigration will simply fail at this point, and the subject might simply say, yes, I still know that there is something representing a red strawberry there, and I can possibly behave the same, but it isn't phenomenally like the original t all.
If we are looking at a strawberry patch when this transmigration operation is performed, there is something in our brain that has green phenomenal properties representing the leaves and something with red phenomenal properties representing the strawberries. This red and green knowledge in our conscious world of awareness is what enables us to pick out the strawberries from amongst the leaves.
There are things in nature that have the phenomenal properties of red and green. There is a notion of 'effing' the ineffable where you can configure something with a different phenomenal property (lets assume something no human has experienced and call it "gred"), and unify it into a world that is already aware of red and green. When you throw the configuration switch in this now properly augmented and unified mind, the conscious entity may reply: "Oh THAT is what "gred" is like".
The entire conscious system that unifies all this red and green into one phenomenal 3D world obviously uses the difference between red and green to represent and distinguish between the red strawberries and the green leaves. The fact that we know the difference between red and green, and the fact that it is aware of them at the same time in a unified effing way, is what gives us the ability to pick out the strawberries from amongst the leaves.
Typical virtual simulations can accomplish representations using abstracted numbers, say 1 to represent red, and 0 to represent green, but there is no known way to unify any such distinguishable representations so the entire system is aware of an entire scene at the same time as we are. This well known problem is sometimes referred to as the binding problem
Of course neurons have many properties individual bits, or even registers of bits in computers lack, that could possibly accomplish the unification of all this stuff into one world of awareness. The fact that when a neuron fires it communicates near simultaneously to many downstream neurons at the same time, and the way neurons fire in synchronous wave patters likely has something to do with the way things are 'bound' to produce our effable, unified 3D world of red and green.
Any virtual system resulting from transmigration will have to accomplish this binding in some yet to be discovered way, so that the resulting 'mind' can be aware of the leaves and the strawberries, at the same time, in a unified or 'bound together' world that is indistinguishable, behaviorally at least, from the phenomenally conscious worlds we know so phenomenally well.
If there is a set of neurons responsible for one spot or voxel (3D volume pixel) of red on the surface of our knowledge of the strawberry in our unified and effable conscious 3D world that is our awareness, let us imagine what could happen is we transmigrate this set of neurons.
When this set of neurons is simulated, the simulation must communicate into the remaining effably unified world of awareness so that the world can continue to distinguish this red voxel from the green voxels representing the leaves in the unified conscious space.
But of course, if there is some unique state of matter for which only it has this 'red' phenomenal property, as would be true in this theory, any abstract 1 or 0, whether such is represented by silicone or any other abstract representation will not be able to accomplish this. Any such abstract simulation could in no way fill the hole where the phenomenally red spot once was.
It might be theoretically possible to create an abstracted world of unified awareness using something like ones and zeros, enabling a similarly unified and aware system to pick out the 1 colored strawberries from amongst the 0 colored leaves. Theoretically such a system could behave like us as it picked the strawberries from amongst the leaves.
To enable transmigration, perhaps a binding way could be found to unify the beginnings of a simulated world of ones and zeroes (or whatever is used as in an inverted qualia and such scenario) with an existing unified world of red and green. But even with this, when the 'effing' switch is thrown between the two, the ones and zeros in this effable unified world would clearly and obviously be different than the red and green.
One might be tempted to argue that you could skip this one at a time unification problem, and jump, entirely, to a complete system representing a strawberry patch with a unified world of ones and zeros (or any other different representation), avoiding the need for the system to distinguish the ones and zeros from the red and green. One could wire such a system to simply answer a question like: "What is red like for you?" with an answer like 'red'.
But of course, any such answer would be a lie. And given such 'effing' abilities, and awareness of what it is that reliably does and does not have red phenomenal properties, and how they are different from ones and zeros, we would be able to prove and effably demonstrate that such was a lie.
Here we have shown one possible theory, which if true, indicates there will be clear problems when attempting any transmigration of a unified and phenomenally conscious mind. This clearly demonstrates the fallacy of using this transmigration argument to claim abstracted simulations would have the same phenomenal properties that real unified phenomenal minds have.