Why You Read Something and Still Can't Explain It

Why can I read something and still not be able to explain it?

Because reading completes only half the job. Following the sentences builds the textbase, the ideas as written, and skilled readers do that with almost no effort or awareness. Explaining needs the situation model, which forms only when you connect the text to what you already know and work out what follows.

Reading researchers describe text memory in layers: the wording, the ideas as the text states them, and the situation the text describes, assembled from those ideas plus your own knowledge. In their account of Kintsch's construction and integration theory, Donna Caccamise and Lynn Snyder are blunt: creating the textbase happens with little awareness and is essentially effortless in skilled readers, while deeper comprehension happens only when a reader also builds a situation model, an effortful, almost problem-solving process. Their article is free to read in full.

In a 2012 Reddit thread asking for a name for the phenomenon, one commenter wrote that they can grasp most things well enough to fumble around with them, but would ultimately fail to explain them.

Why does restating the page feel like real understanding?

Because a coherent page hands you the first two layers at no cost. The words are printed and the ideas are laid out in order, so restating the text costs you nothing. Clear writing never forces the inferential work that builds the third layer, and the smoothness of that free pass feels like mastery.

The model assumes that a text is rarely completely explicit. Readers have to bridge gaps: resolving a pronoun, connecting an example to the broader point, inferring what the writer left unsaid. Good writers close most of those gaps for you, so the page feels whole and nothing signals that the third layer was never assembled. The problem is not effort. It is that nothing demanded any. A reader on Reddit described knowing what a word means inside a sentence, then being unable to define it when asked, and wondering whether that was unusual. It is not unusual.

How do I tell whether I actually understood something or just followed it?

Ask an applied question instead of a recital one. What did it say can be answered from the surface of the text. What would change if this were true, or what does it imply for next quarter, cannot. Research on retrieval instructions shows that the two kinds of probe pull up different layers of memory.

In a 2022 study in the American Journal of Psychology, Stephen Chronister, Andrea Tamplin and Gabriel Radvansky gave readers one of two instructions. Verbatim instructions asked whether a sentence had actually appeared in the text. Consistency instructions asked whether a sentence was consistent with what had been read. Verbatim instructions drew on the surface form and the textbase, consistency instructions on the situation model, and the split was still there a week later. Their paper is available as a full text PDF. So the self-test you have been using is unreliable: a fluent speaker can restate the textbase and sound finished. What would this change cannot be answered that way.

Can I understand something and still not be able to explain it?

Sometimes yes. Explaining is its own skill, separate from knowing, and turning an experience into words is an active process that does not always read out cleanly. A blank explanation is a real signal, but on its own it does not prove that you never understood the material.

People argue about exactly this. In the same thread, one commenter insisted that failing to explain an idea is what not understanding looks like, while another pointed out that you can understand something well and simply lack the speaking ability to explain it. Procedural knowledge is knowing how to do a thing, like swinging a golf club, and it can exist with no ability to describe the steps. Work on verbal overshadowing adds a second mechanism: describing an experience is an active process, not a transparent readout of what is stored. Brown and colleagues on when words hurt is open access. So a visible failure to explain is not proof of an absent model, only a reason to run the applied question.

What should I do in the ten minutes after I read something that matters?

Build one bridge before you move on. Write a single sentence about what the idea would change, in your own words, and say it out loud once. That is the applied question and the generation step in one move, and it takes less than a minute.

After a section that matters, answer what this would change if it were true. Before you turn the page, predict what comes next and check yourself. Before a meeting or a comment, say your one sentence out loud. Restating is the part reading does for you. Bridging is the part that makes an idea survive the next morning. You have probably heard the meeting line that if you cannot explain something simply, you do not understand it. People attribute it to Einstein, who never said it, and Wikiquote lists it among his misattributed quotes. It also tests the wrong thing: simplicity measures how well you package an idea, not whether the model is there. What Absorb scores is production, in a short daily round where you either produce the answer or you do not. See Absorb on the App Store if you want that round in your day.