Yes you are correct that the original argument in this thread was not a general attack on physicalism (even the issue that I am discussing with Jonathan Edwards on this same thread is not, but that is an attack on a quite widely held physicalist perspective). I think the argument is similar to the bat argument. But a person might think that they could work out the experience the Mark 19 robot would have had with the colour vision cameras in (a functionalist might think they were able to). It then seems to follow if people could, then why not RoboMary. And if RoboMary could, then why could it not be worked out even when only having the black and white cameras in (which is what I was assuming Dennett was claiming). Furthermore if RoboMary could have worked out what it would be like to consciously experience blue, when only having had black and white cameras, then RoboMary would have worked out what it would be like to experience something she had not previously experienced. Which would then bring in doubt the claim that it would be impossible to work out what a bat would consciously experience, just because we lack the experience. Why could we not work it out like RoboMary? Dennett is not stating how RoboMary could work what the experience of blue was like, but seemed to be challenging others to show that she could not. The argument I supplied just showed a simple scenario in which it can be seen that in room 3 there would be no basis to assume the experience being more like your experience of blue or more like your experience of red. You had touched upon that the response could be that the experience would change between red and blue, but remember the robot could not report any difference in experience, and so if it were claimed that in the third room the experience changed from red to blue, then consider the following statement by Dennett:

Thinking in terms of robots is a useful exercise, since it removes the excuse that we don’t yet know enough about brains to say just what is going on that might be relevant, permitting a sort of woolly romanticism about the mysterious powers of brains to cloud our judgment. If materialism is true, it should be possible (“in principle!”) to build a material thing–call it a robot brain–that does what a brain does, and hence instantiates the same theory of experience that we do

But if the robot brain did instantiate the same theory of experience that we do as Dennett suggests and yet it was be unable to respond to the change of experience from red to blue suggested, then that would suggest that we, like the robot, cannot respond to what we consciously experience. Which I think was the point that zombie arguments were suggesting regarding some physicalist theories.  So the thought experiment opens up attacks on different points depending on the response given, which I think is different from the bat argument. 

By the way do you dispute the following:

You can tell that reality is not a zombie universe based on your knowledge that you conscious experience, and you can conclude from that that what you consciously experience is having an influence on your behaviour (you are using it as evidence in your claim that reality is not a zombie universe).

Also regarding zombies, and this is just a pretty unrelated side issue, only being mentioned to avoid the topic changing to whether the idea of them is compatible with physicalism:  they could be imagined in a universe which is different physically, but which follows the same laws of physics, such that any experiment would give the same results. This would just be to conceive of them in a way which is not incompatible with physicalism (no suggestion of two physically identical things having different features).

Regarding your point about the behaviourist or functionalist that claims they do not change the subject. If they are simply stating that if something has certain features (functions in a certain way for example) then it will also have the feature of consciously experiencing, then that is fine, they can still follow the conversation though and they can understand the point in the argument. 

However if they suggest that you did not mean the feature you did by consciously experiencing but other features, then you are in a position to correct them, the same as any other time a person might misunderstand what you meant. For example if you felt like indulging them, you could listen to the features they thought you meant, paraphrase those features back to them, just to allow them to be clear that you understand the features they are talking about, before informing them that you are discussing another feature. They can claim to be ignorant about which feature you are discussing, but why should their ignorance be your problem? Indeed if they were suffering from some type of Emperor's New Clothes Syndrome where they thought it was clever to see it as though the feature was absent you could well be wasting your time while they remained inclined to think it was clever to claim to see it that way. I would be interested in what you make of the conversation I was having with Derek Allen on this thread. Anyway, while I accept that it could be a problem if you were in an academic institution where you might be relying on such people to understand an argument which contained the concept of consciously experiencing, other than that, they are just sidelining themselves by claiming they cannot understand what obvious feature of reality other people were discussing. It seems to me it would be strange for them to think they could join in the discussion while claiming to be ignorant of the feature the people were discussing. They could just be ignored while the conversations are carried on with those that do understand. You can claim they are ignorant of what you are discussing, but they cannot claim the same of you in return regarding what they are discussing (that it entails more than your paraphrasing of the discussion). 

Though, I am not sure that these latter type of behaviourists or functionalists are mainstream. I assume they misunderstood the mainstream position. But that is because I have assumed mainstream behaviourists are of the mind that in a scientific context they should limit themselves to discussing behaviour, because that is what they consider to be the subject matter of science, and therefore in a scientific context the term consciously experiencing should take on a behavioural meaning. In the same way that words can take on different meanings in different contexts. That would be changing the subject though if that meaning were used in response to a philosophical argument which was not using that meaning. And I assume a functionalist would normally advocate functionalism as an explanation of the type of behaviour which gives rise to the feature. And so would not be a denying that they understood what feature they were suggesting certain functional behaviour would give rise to.