if thought truly is entirely deterministic, then it's surely both sufficient and necessary that you could build a machine that, given the state of the universe as input, could fully simulate what your answer to any given question would be

but if you suppose that, then you basically run into an issue very similar to the halting problem

you put your subject in the room with your magic machine, tell them to disagree with whatever the machine spits out, then tell the magic machine to predict what they're going to say after they've been told the result of said prediction

whatever the machine spits out, there's nothing stopping your subject from just disagreeing