Americans are not good readers. Many blame the ubiquity of digital media. We’re too busy on Snapchat to read, or perhaps internet skimming has made us incapable of reading serious prose. But Americans’ trouble with reading predates digital technologies. The problem is not bad reading habits engendered by smartphones, but bad education habits engendered by a misunderstanding of how the mind reads.

Just how bad is our reading problem? The last National Assessment of Adult Literacy from 2003 is a bit dated, but it offers a picture of Americans’ ability to read in everyday situations: using an almanac to find a particular fact, for example, or explaining the meaning of a metaphor used in a story. Of those who finished high school but did not continue their education, 13 percent could not perform simple tasks like these. When things got more complex — in comparing two newspaper editorials with different interpretations of scientific evidence or examining a table to evaluate credit card offers — 95 percent failed.

There’s no reason to think things have gotten better. Scores for high school seniors on the National Assessment of Education Progress reading test haven’t improved in 30 years.

Many of these poor readers can sound out words from print, so in that sense, they can read. Yet they are functionally illiterate — they comprehend very little of what they can sound out. So what does comprehension require? Broad vocabulary, obviously. Equally important, but more subtle, is the role played by factual knowledge.

All prose has factual gaps that must be filled by the reader. Consider “I promised not to play with it, but Mom still wouldn’t let me bring my Rubik’s Cube to the library.” The author has omitted three facts vital to comprehension: you must be quiet in a library; Rubik’s Cubes make noise; kids don’t resist tempting toys very well. If you don’t know these facts, you might understand the literal meaning of the sentence, but you’ll miss why Mom forbade the toy in the library.

Knowledge also provides context. For example, the literal meaning of last year’s celebrated fake-news headline, “Pope Francis Shocks World, Endorses Donald Trump for President,” is unambiguous — no gap-filling is needed. But the sentence carries a different implication if you know anything about the public (and private) positions of the men involved, or you’re aware that no pope has ever endorsed a presidential candidate.

You might think, then, that authors should include all the information needed to understand what they write. Just tell us that libraries are quiet. But those details would make prose long and tedious for readers who already know the information. “Write for your audience” means, in part, gambling on what they know.

These examples help us understand why readers might decode well but score poorly on a test; they lack the knowledge the writer assumed in the audience. But if a text concerned a familiar topic, habitually poor readers ought to read like good readers.

In one experiment, third graders — some identified by a reading test as good readers, some as poor — were asked to read a passage about soccer. The poor readers who knew a lot about soccer were three times as likely to make accurate inferences about the passage as the good readers who didn’t know much about the game.

That implies that students who score well on reading tests are those with broad knowledge; they usually know at least a little about the topics of the passages on the test. One experiment tested 11th graders’ general knowledge with questions from science (“pneumonia affects which part of the body?”), history (“which American president resigned because of the Watergate scandal?”), as well as the arts, civics, geography, athletics and literature. Scores on this general knowledge test were highly associated with reading test scores.

Current education practices show that reading comprehension is misunderstood. It’s treated like a general skill that can be applied with equal success to all texts. Rather, comprehension is intimately intertwined with knowledge. That suggests three significant changes in schooling.

First, it points to decreasing the time spent on literacy instruction in early grades. Third-graders spend 56 percent of their time on literacy activities but 6 percent each on science and social studies. This disproportionate emphasis on literacy backfires in later grades, when children’s lack of subject matter knowledge impedes comprehension. Another positive step would be to use high-information texts in early elementary grades. Historically, they have been light in content.

Second, understanding the importance of knowledge to reading ought to make us think differently about year-end standardized tests. If a child has studied New Zealand, she ought to be good at reading and thinking about passages on New Zealand. Why test her reading with a passage about spiders, or the Titanic? If topics are random, the test weights knowledge learned outside the classroom — knowledge that wealthy children have greater opportunity to pick up.

Third, the systematic building of knowledge must be a priority in curriculum design. The Common Core Standards for reading specify nearly nothing by way of content that children are supposed to know — the document valorizes reading skills. State officials should go beyond the Common Core Standards by writing content-rich grade-level standards and supporting district personnel in writing curriculums to help students meet the standards. That’s what Massachusetts did in the 1990s to become the nation’s education leader. Louisiana has recently taken this approach, and early results are encouraging.

Don’t blame the internet, or smartphones, or fake news for Americans’ poor reading. Blame ignorance. Turning the tide will require profound changes in how reading is taught, in standardized testing and in school curriculums. Underlying all these changes must be a better understanding of how the mind comprehends what it reads.