The complete but abridged case against functionalism (and functionalist reductionism in particular) has five parts and goes as follows:
1. The Binding Problem.If all you see is computation, there is no principled mechanism to determine which parts of computation form unified entities, or alternatively put, no way to draw boundaries around computational processes. There's no way to explain why there is one consciousness in the human brain observing one unified visual field, as supposed to 10000 consciousnesses each observing one pixel.
2. The Interpretation problemIf all you see is computation, there is no way to decide what is being computed. Consider the Glider Gun, for example:
A thing that produces the 5-pixel "glider" thingies? Or a machine that counts up forever? Or a calculator that computes 2+3=5 once (and then the rest is ill-defined)? Or that computes 5*5=25? Each of these interpretations, and infinitely more, are valid given the appropriate input/output mapping (E.g., if the output is the number of pixels in the gliders, it outputs 5.)
3. The Abstraction ProblemGoL inherently consists of non-material computational units (cells). But in the real world, there is no ground truth as to what even constitutes a Bit.
Could a single person pushing rocks simulate the universe (because they implement a Turing Machine)? Is the united states phenomenally conscious (because people/groups implement computational units)? Are fictional characters phenomenally conscious? Companies? If you simulate a Turing Machine on a computer, do the TM's steps or the computer's steps count? In a normal computer, who decides that "magnetic charge" and "no magnetic charge" correspond to 1 and 0, as supposed to "high magnetic charge" and "all but high magnetic charge"? No principled way to answer any of these questions.
4. The Plausibility ArgumentIf you forget all the reasons why you think consciousness is tied to intelligence and whatnot, the claim that this made-up concept "Bit" is the source of consciousness, rather than matter, seems not a-priori plausible.
5. The AlternativeI think in practice, people like Functionalism
because of groupthink because there appear to be knock-down arguments against all alternatives. And I agree with almost all of them, except that the knock-down argument against panpsychism is
Scan a human brain. Build a neuron-level simulation. It will run exactly the same computational steps as the original. If the original wrote papers about consciousness, the simulation will write papers about consciousness. It's absurd to claim that the second one does this without being conscious.
Which I would agree with UNLESS computations in the brain are substrate dependent, in which case the simulation *doesn't* have the same input/output behavior