My hunch, for what it’s worth, is that it is indeed “the physicality of the printed page” that makes a significant difference — in a couple of specific senses.
First of all, the stability of the text on a printed page allows us (as most readers know) to have visual memories of where passages are located: we see the page quadratically, as it were, divided into upper left, lower left, upper right, and lower right. This has mnemonic value.
Second, the three-dimensionality of a book allows us to connect certain passages with places in the book: when we’re near the beginning of a book, we’re getting haptic confirmation of that through the thinness on one side and thickness on the other, and as we progress in our reading the object in our hands is continually providing us with information that supplements what’s happening on the page.
A codex is then an informationally richer environment than an e-reader.
This is is exactly why I use physical textbooks vs pdf versions. With a physical book, I can remember where in the text particular principles were introduced. If I could just get a search function to work on a physical book….