But if by behavior we don't mean movements by the body, and instead the whole state of the brain (the entire thinking process), then the theory is again trivial

But if by behavior we don't mean movements by the body, and instead the whole state of the brain (the entire thinking process), then the theory is again trivial and useless (because it doesn't explain anything). Indeed I don't claim to have read too much about it, but I think the main stream of behaviorism is literally that the functional response (i.e. black box response) is essentially all that matters. I'd be happy to know if there's more nuance than what appears. I have a number of qualms about functional explanation of consciousness (which goes back to the Turing Test, probably earlier). I've seen outright denials that a representation of what is currently on your mind (the "mental theater"), often associated with consciousness, and indeed all similar abstractions we often use referring to our own minds, as "illusions" -- the brain is seen (by I believe at least a few currents) as a single monolithic black box function without a usefully interpretable internal state. Anyone that has in fact experience programming computers will know there usually distinct caches, and clearly distinct subsystems that process information alternatively or in parallel, and algorithms operate recursively or iteratively over the memory until reaching a solution. I don't see why those insights shouldn't hold value for explaining consciousness, and perhaps explaining that even if an algorithm (e.g. a simple machine learning system) is indistinguishable from a [median] human in one or all tasks, then it is still not necessarily conscious, because the details of its inner workings lack certain features of a truly conscious being. For example, its internal representation of emotions may not be equivalent that of a human, such that it does not actually "experience" emotions, but because our own capabilities of verbally communicating emotions are quite limited, it may well beat the median human at this ability after utilizing gargantuan training sets. So you can see themes of data efficiency, internal representation, etc emerge that are completely left out of behaviorism and related theories. I intend to write a little about this (in an article), but there is probably at least a few academics exploring this more complicated, and complete, understanding.