Projecting the ontology of your (flawed) internal representations onto reality is a bad idea. "Doing a Dennet" is also not dealt with, except by incredulity.

It's a fact that the individual shades of color exist, however it is that we group them - and your ontology must contain them, if it pretends to completeness.

This is simply not the case. The fact that we can compare two stimuli more accurately than we can identify a stimuli merely means that internally we represent reality with lesser fidelity than our senses theoretically can achieve. On a reductionist view at most you've established "greater than" and "round to nearest" are implemented in neurons. You do not need to have colour.

Let's unpack "blueness". It's a property we ascribe to objects, yet it's trivial to "concieve of" blueness independent of an object. Neurologically, we process colour, motion, edge finding and so in in parallel; the linking of them together occurs at a higher level. Furthermore the brain fakes much of the data, giving the perception of colour vision, for example, in regions of the visual field where no ability to discriminate colour exists, and cases of blindness with continued concious perception of colour.

Brains compress input extensively; it would be crass to worry about the motion of every spot on a leopard separately - block them up as a single leopard. Asserting that the world must fit with our hallucination of reality lets you see things that are marginally visible, and get by with far worse sensory apparatus than needed. Cue optical illusions: this, this and this, for example. Individual shades don't exist as you want them to.

It is absurdly clear that the map your brain makes does not correspond to either the territory of your direct sense perception (at the retina) or reality. On precisely what basis do you assert to project from the ontology of a bad map to the territory?

"Blue" is a referent to properties of internal representations, which is translatable across multiple instances of primate brains. You say "X is blue", and I can check my internal representation of X to see whether I would categorise it as "blue". This does not require "blue" to be fundamental in ontology. There isn't a "blue thing" in physics, nor should there be. "Blue" existing means simply that there are things which this block of wetware puts in some equivalence class.

Lets move on to computation:

But if the "computational state" of a physical object is an observer-dependent attribution rather than an intrinsic property, then how can my thoughts be brain states?

Again, you seem project from an internal map of your own brain to the territory. Simply because I can look at a computer at multiple levels, say: Starting Excel, API calls, Machine instructions, microcode, functional units on the CPU, adders/multipliers/whatever on the CPU, logic gates, transistors, current flows or probability masses in the field of electrons, does not in principle invalidate any of the above views as correct views of an operating computer. The observer dependence isn't an issue if (modulo translation/equivalence classes for abstraction between languages) they all give the same function or behaviour. You can block things up as many low level behaviours or a smaller number of high level ones; this doesn't invalidate a computational view. What is the computation implemented by starting Excel? What details do you care about? It doesn't matter to a functionalist, as the computations are equivalent, albeit in different languages or formalisms.

The critique of aboutness is similar to your issues over colour. You percieve "X is about Y" and thus assume it to be ontologically fundamental. Semantic content is a compressed and inaccurate rendition of low level states: Useful for communicating and processing if you don't care about the details. Indeed the only reason we care about this kind of semantics is that our own wetware implements theory-of-mind directly. Good idea for predicting cognitive agents; not neccessarily a true statement about the world. The "Y" that "X" is "about" is another contraction - an infered property of a model.

"Time" is as flexible as your neural architecture wants it to be. Causality is a good idea, for Darwinian reasons, but people's perception of the flow of time is adjustable. I will point out that your senses imply strongly that the world is a 2D surface. Have you ever been able to see behind an object without moving your head? I haven't either, therefor clearly this 3D stuff is bunkum - the world is a flat plane and I directly percieve part of one side of it. Ditto time. Causality limits the state of a cognitive thing to be dependent on its previous states and its light cone at this point in space-time, and you percieve time to flow because you can remember previous brain states, and depending on them (compressed somewhat) is good for survival.

And now for unity of conciousness. It isn't unitary. Multiple personality, dissociative disorders, blindsight, sleepwalking, alien hand, need I go on? I percieve my own representation of reality to be unitary; I know for a fact that it's half made up. You claim that the individual issues "just can't" be the whole story. Why? Personal incredulity isn't an argument. The brain in the skull you call yours isn't just running a single cognitive entity. You move before even realising "you" were going to; you are unconcious of breathing until you decide to be. Why is a unitary conciousness fundamental? Why isn't it just a shortcut to approximate "you" and others in planning the future and figuring out the present?