Why Do I Read the Same Sentence Over and Over — Eye Regressions
You reach the end of a sentence and your eyes slide back to the start of it. You read it again. Somewhere in the middle you feel the meaning slip, so you go back a third time. By the time it finally lands, you are mildly irritated with yourself, and the next sentence is already waiting.
Maybe it happens with contracts and dense emails. Maybe it happens with novels you genuinely want to read. Either way, the question underneath is the same: is something wrong with how I read?
The short answer, from four decades of eye-tracking research, is that looking back is a normal and useful part of reading. The longer answer is more interesting. How often you look back, and why, says a lot about how your working memory, attention, and stress systems are handling the text in front of you.
What are regressions in reading
Reading feels like a smooth glide from left to right. It is not. Your eyes move in a series of rapid jumps called saccades, pausing on words in brief fixations. Most of those jumps go forward. A meaningful minority go backwards, and reading researchers call those backward jumps regressions.
How common are they? Estimates vary with the reader and the text, but studies tracing back to the late Keith Rayner, one of the founders of modern eye-tracking research, put regressions at roughly 10 to 15 percent of eye movements in skilled adult readers, with ranges of 5 to 20 percent reported across studies. Some readers regress far more than that and still understand perfectly well.
Many regressions are tiny, a hop back of a word or two that you never consciously notice. The ones you do notice, where you deliberately go back to the start of the sentence, are the far end of the same spectrum. Rereading is not a glitch in an otherwise linear process. It is built into the process.
Do regressions actually help you understand
For decades, speed-reading courses taught the opposite. Regressions were framed as a bad habit to be trained out, and in the 2010s a wave of apps promised to remove them entirely by flashing one word at a time in the same spot on the screen.
Elizabeth Schotter, Randy Tran, and Keith Rayner tested that promise directly in a 2014 study in Psychological Science with a memorable title: "Don't Believe What You Read (Only Once)." Forty university students read sentences normally, or with each word replaced by a row of Xs as soon as their eyes moved past it. Going back was still physically possible, but there was nothing left to see.
Comprehension dropped when rereading was blocked, for straightforward sentences as well as tricky ones. In normal reading, by contrast, readers understood just as well on trials where they regressed as on trials where they did not. That pattern suggests looking back is targeted: readers go back when they need to, and it works.
"When readers cannot backtrack and get more information from words and phrases, their comprehension of the text is impaired." — Elizabeth Schotter
A larger 2016 review in Psychological Science in the Public Interest, led by Rayner with Schotter, Michael Masson, Mary Potter, and Rebecca Treiman, reached the same conclusion about speed reading in general. Eye movements account for no more than about 10 percent of reading time, so eliminating them saves surprisingly little. What limits reading speed is language processing, not the eyes.
"The available scientific evidence demonstrates that there is a trade-off between speed and accuracy." — Elizabeth Schotter
So if you reread, the first thing to know is that you are using a tool, not exhibiting a flaw. The useful question is why you are reaching for that tool more often than you would like.
Why sentences send your eyes backwards
Some sentences simply demand a second look. Linguists call them garden-path sentences, because they lead you confidently down one interpretation before revealing it was the wrong one. The classic example is "The old man the boat." Most readers parse "the old man" as a noun phrase, hit a dead end, and have to go back to realise that "man" is the verb.
Lyn Frazier and Keith Rayner's 1982 eye-tracking study of sentences like these, published in Cognitive Psychology, found that readers often sent their eyes back to the point of confusion rather than restarting the whole sentence. Later work has shown that this selective reanalysis is not universal. Many readers reread more broadly, especially when they are not sure exactly where the misunderstanding began.
Everyday writing is full of milder versions of the same problem: long sentences with the main verb buried at the end, stacked qualifications, ambiguous pronouns, technical vocabulary. When the text is badly built, frequent rereading says more about the writing than about the reader.
How working memory drives rereading
When the sentence is clear and you still keep going back, the likeliest bottleneck is memory and sequencing, the dimension that covers working memory.
To understand a sentence, you have to hold its early parts in mind while you process the later parts, then bind them into a single meaning. Meredyth Daneman and Patricia Carpenter showed in 1980 that a task measuring exactly this juggling, reading sentences while remembering their final words, predicted reading comprehension far better than simple memory span. Later meta-analytic work confirmed that complex working memory measures are among the stronger predictors of comprehension in adults.
If the subject of a sentence has faded from working memory by the time you reach its verb, the sentence does not cohere. Your eyes go back to retrieve the piece that dropped out. Seen this way, a regression is an external memory lookup: the page is holding information your internal workspace could not.
Eye-tracking research supports this link, with some nuance. A 2014 study in Frontiers in Human Neuroscience found that working memory capacity partly predicted how efficiently readers aimed long-range regressions back to earlier parts of the text. Other studies show that reading speed and vocabulary account for much of the remaining variation. Working memory is not the whole story, but it is a large part of it.
There is a twist. In a classic 1983 study, Richard Olson, Reinhold Kliegl, and Brian Davidson described some readers as "explorers" who looked back and forth along the line, and others as "plodders" who rarely regressed. The explorers were not the weaker readers. Frequent looking back can be an active, strategic way of building meaning. If you want the fuller picture of why information evaporates mid-page, we cover it in why you forget what you just read.
Is rereading a sign of ADHD
Rereading by itself is not diagnostic of anything. But there is growing evidence that it looks different in readers whose attention and rhythm dimension runs on a different pattern.
Pnina Stern, Lilach Shalev, and colleagues tracked the eyes of adults with and without ADHD as they read continuous text, publishing in the Journal of Attention Disorders in 2024. The ADHD group took significantly longer to read, and occasions where a word was reread more than twice were much more frequent among them. The authors describe this repeated rereading as a marker of inattentive reading.
That fits a mechanism Erik Reichle, Andrew Reineberg, and Jonathan Schooler captured in 2010. Readers pressed a key whenever they caught themselves zoning out while an eye tracker recorded them. During these episodes of "mindless reading," fixations got longer and stopped responding to the properties of the words, and eye movements became especially erratic just before readers noticed. The eyes kept moving, but language processing had quietly left. The repeated reread is what happens when you come back and realise you missed it.
A 2025 study in the Journal of the International Neuropsychological Society adds an intriguing complication. Moussaoui, Niemeier, and colleagues found that young adults with ADHD understood text nearly 13 percent better when words were presented one at a time in a single location, which removes the need for eye movements, while control readers did worse in the same format. It is one small study of university students, and it does not overturn Schotter's findings. But it suggests the best reading format may depend on your cognitive profile rather than being universal.
Why anxiety makes you read the same line again
The third dimension that often sits behind rereading is emotional regulation, and it can operate in two quite different ways.
The first is capacity. Michael Eysenck's attentional control theory proposes that anxiety takes up part of working memory with worry, leaving less space for the task. Eye-tracking studies of induced stress and trait anxiety, mostly small, report longer reading times and more and larger regressions in anxious readers. You reread not because the sentence is hard, but because part of your mind was rehearsing tomorrow's meeting while your eyes crossed it.
The second is doubt. For some people, rereading stops being about comprehension and becomes about certainty: going back to make sure you really read every word, really understood, did not miss something important. Clinicians who treat obsessive-compulsive disorder recognise compulsive rereading as a form of checking.
Research by Marcel van den Hout and Merel Kindt in the early 2000s showed something counterintuitive about checking. Repeated checking does not improve confidence in memory. It erodes it, because each repetition makes the original memory feel less vivid and distinct. The more you reread to feel sure, the less sure you feel. If rereading feels driven by anxiety rather than confusion, that pattern is worth discussing with a qualified clinician.
Does dyslexia cause more regressions
Readers with dyslexia do show more fixations and more regressions on average. For a while in the 1980s, some researchers argued that faulty eye control might cause dyslexia, and that argument still circulates online.
Rayner's work, and most of the field since, points the other way. When dyslexic readers are given non-reading tasks, their eye movements largely look typical. The extra regressions mainly reflect the extra effort of decoding words, which ties back to phonemic processing and visual processing rather than to the eye muscles. A minority of studies have reported subtle oculomotor differences in some dyslexic readers, and that debate is not entirely closed.
The practical point is that eye-movement training is not a treatment for dyslexia. If decoding is effortful, the regressions are a symptom, and the support that works targets the language system. You can read more about where that effort comes from in why some people read slowly for reasons beyond dyslexia.
Is rereading a good way to study
Here is the paradox. Looking back within a sentence helps comprehension. Rereading whole passages to learn them is one of the weakest study strategies available.
In a major 2013 review in Psychological Science in the Public Interest, John Dunlosky, Katherine Rawson, Elizabeth Marsh, Mitchell Nathan, and Daniel Willingham rated rereading as "low utility" for learning, alongside highlighting. Practice testing and spacing out study sessions earned the only "high utility" ratings. Rereading feels productive because the text becomes familiar, but familiarity is not the same as being able to recall it.
So the regressions your eyes make while reading are doing real work. The habit of reading a chapter three times before an exam is mostly not.
What you can do when you keep rereading
Start by noticing which kind of rereading you are doing. A useful test is to ask, in the moment, what you went back for. If you can name it, such as the subject, the number, or the clause you lost, you are probably using regressions as intended. If you cannot, attention or anxiety is the more likely driver.
For working memory pressure, reduce the load on the internal workspace. Break long sentences down by jotting the subject and main claim in the margin. Read difficult material when you are rested. Paraphrase each paragraph in one line before moving on, which turns a fragile memory trace into something more durable.
For attentional drift, change the rhythm rather than forcing harder concentration. Shorter reading blocks, reading aloud, following text with a finger or cursor, or listening while you read can all give attention an external anchor. Given the 2025 ADHD findings, it may be worth experimenting with one-word-at-a-time apps for short texts, while keeping in mind that most readers understand less in that format.
For anxiety-driven rereading, the goal is to tolerate a little uncertainty. Set a rule of one reread per sentence, then move on and see whether meaning recovers from context. It usually does. If the urge to reread feels compulsive and distressing, evidence-based therapy such as exposure and response prevention is designed for exactly that loop.
It can also help to understand how your own dimensions combine, since the same rereading habit has very different roots in different people. CognitionType offers a useful way to explore your own cognitive profile across dimensions like memory and sequencing, attention and rhythm, and emotional regulation. It works best alongside, not instead of, a formal assessment when you suspect a specific condition.
When frequent rereading is worth investigating
Occasional rereading is universal. It becomes worth a closer look when it is constant across all kinds of text, when it makes reading so slow that it affects work or study, when it comes with long-standing difficulty decoding or spelling, or when it is driven by distress rather than confusion.
In those cases, an educational psychologist, neuropsychologist, or clinician can separate working memory, attention, decoding, and anxiety in ways self-observation cannot. The goal is not to stop your eyes going back. It is to understand why they go back so often, so that the support you choose fits the mind doing the reading.
CognitionType is an informational assessment, not a clinical diagnosis. If you suspect ADHD, dyslexia, an anxiety disorder, or another specific condition, we encourage you to seek a formal evaluation from a qualified professional.