Consider the following sentence:
“I was never certain that I didn’t want to go to the cinema”.
You can make sense of it if you sit and think about it for a while but it is pretty confusing. This is why writing guides suggest avoiding the use of the ‘double negative’: A sentence or phrase with two ‘not’ terms which, in this case, are the ‘never’ and the ‘didn’t’.
It could simply be the case that we find the sudden reversal of meaning unusual and this is what trips us up.
But it might be a cognitive load effect. Many sentences, even complex ones, consist of clauses linking just two or three ideas. By adding a double-negative, we have not only added another item but this item also interacts with the items already present by changing their meanings. So we now need to attend to the new relationships this creates. Together, this overloads working memory.
John Sweller is one of the greatest proponents of this idea of ‘element interactivity’ as a measure of complexity but it has been criticised by others for not being wholly quantifiable and for not yet being manipulated with experiments (although studies have begun to attempt to do this).
I suspect that element interactivity might prove to be a useful model.