A puzzle about fiction and belief

Steven Fielding's new book on the representation of British politics in fiction contains the following from Geoffrey Howe, Cabinet Minister whose resignation precipitated the fall of Mrs Thatcher. He had been watching a dramatization of these events on TV:

At almost every moment when my actions, my words, were being depicted, I was conscious of serious, no doubt unintentional inaccuracies. Literally nothing was quite right. Yet for all those sequences when I was not on screen, disbelief was largely suspended. The talking moving picture is a compelling witness. ‘So that’s why George – or Peter or whoever – ‘did that’, I found myself thinking time and again. Beguilingly, the cameras appeared to be telling the truth, except where I positively knew them to be inventive and false (Conflict of Loyalty, (London, Macmillan, 1994), p.683, quoted in Stephen Fielding, A State of Play (London, Bloomsbury, 2014), pp.15-16)

It is striking that he of all people, knowing how unreliable the representation was concerning his own actions, was prone to  believe what was represented of others. Would the same effect have been produced by a novelistic fiction? That seems less likely, though there is plenty of evidence of a tendency to believe what is implied by written fiction also. Does visually encoded representation slide more easily past our sub-personal defences? For all that parsing may be a reflex, processing highly realistic visual images seems to be a lot easier than processing language, perhaps because of the heavy burden of pragmatic inference required by the latter. But then there ought to be more resources left to apply epistemic vigilance in the visual case. The explanation can't be that "seeing is believing": Howe had no tendency to believe what he saw himself represented as doing.