This argument certainly applies to reading advanced mathematics texts. (Cross reference: While most advanced mathematicians can read somewhat rapidly through advanced texts very easily following the trail of breadcrumbs the author has laid, out, the author usually takes for granted that the reader has some level of prerequisites to make jumping from one idea to the next possible without including every single stepping stone along the way. Those found without the sometimes unstated required background find themselves more quickly lost and need to spend far more time to be able to jump easily to the next thought.
In Pinker’s framework, advanced mathematics texts do have the requisite coherence relations, it’s just that the distance between the relations are less coherent to some than others (who have more background.)