The Science of Eye Movements in Reading

When you read a sentence, you feel as if your eyes glide smoothly across the words. They don't. Your eyes jump, pause, jump again — in a series of rapid leaps and still moments that happen entirely outside your conscious awareness. The science of these movements, built across more than a century of research, explains nearly everything about reading speed, comprehension, and what tools like RSVP actually do.

What really happens when you read

The basic mechanics were established in the late 19th century and refined throughout the 20th. Here is what actually occurs during a typical line of text:

Fixations. Your eyes land on a word and stay still for roughly 200–250 milliseconds. During this pause, visual information is extracted. You do not read during movement — only during stillness.

Saccades. Between fixations, your eyes jump rapidly. A typical forward saccade covers 7–9 letter spaces and takes 20–35 ms. Information is suppressed during saccades; you are effectively blind during each jump (Rayner, 1998).

Regressions. About 10–15% of all eye movements go backward — returning to earlier text. These are not errors. Research shows regressions correlate with comprehension: they occur when the brain detects a processing failure and directs the eyes back to resolve it (Schotter, Tran, & Rayner, 2014).

Return sweeps. At the end of each line, a large leftward saccade moves the eyes to the beginning of the next line. These return sweeps sometimes undershoot, landing mid-word rather than at the line start.

Fixation duration tells you how hard your brain is working

Fixation duration is not constant. Your eyes pause longer on:

• Unfamiliar or low-frequency words — words you encounter rarely require longer lexical access
• Long words — more letters to process, often multiple fixations on a single word
• Syntactically complex passages — sentences where the grammatical structure is harder to parse
• Semantically dense material — philosophy, technical writing, legal prose

Skilled readers have shorter average fixation durations than developing readers — not because they rush, but because their more robust vocabulary and pattern recognition means lexical access happens faster (Rayner, Schotter, Masson, Potter, & Treiman, 2016).

If you want to read faster without losing comprehension, fixation duration is one key lever. Building vocabulary and familiarity with a domain is one of the most reliable ways to shorten fixation duration naturally.

The perceptual span: how much you see at once

Your fovea — the central 2° of your visual field — handles detailed processing. This covers roughly 7–8 characters. Everything else is processed at lower resolution.

But you extract useful reading information from a wider region than just the fovea. Research using the "moving window" paradigm — where text outside a defined window is masked — established the perceptual span: roughly 3–4 characters to the left and 14–15 characters to the right of your fixation point (Rayner, 1998).

Within this span, parafoveal vision gives you preview information about upcoming words. Before your eyes actually land on a word, you have already partially processed its length, initial letters, and sometimes its phonological form. This preview is crucial for reading fluency — it lets you plan the next saccade accurately and begin processing the next word before you fixate on it.

This is also why RSVP speeds above roughly 400 WPM tend to hurt comprehension. RSVP presents words at a fixed central point, which eliminates parafoveal preview entirely. At moderate speeds, many readers can compensate; at higher speeds, the loss of preview creates a processing bottleneck. This is covered in detail in our post on RSVP versus traditional reading.

Skilled vs novice readers: what the eyes reveal

Eye tracking studies reveal several consistent differences between fast, skilled readers and slower, developing readers:

Skilled readers skip more words — they do not fixate on every word in a line. Function words (the, a, of, in) are often skipped because parafoveal processing has already extracted sufficient information before the eyes arrive. This is one component of what is sometimes mislabelled "peripheral reading" in commercial speed reading courses.

What this means for reading speed

The time your eyes spend in a line of text breaks down roughly as follows:

• Fixation time: 200–250 ms per fixation × number of fixations
• Saccade time: 20–35 ms per saccade (unavoidable)
• Return sweep time: ~50 ms per line

Reading faster means reducing one or more of these. There are three legitimate routes:

1. Reduce fixation duration — through vocabulary building, domain knowledge, and practice with the register
2. Reduce the number of fixations — by processing larger units per fixation (this is the basis of word chunking)
3. Eliminate saccades entirely — which is what RSVP does

RSVP collapses categories 1 and 3 differently from traditional reading. The time saved from saccades is real; the cost from lost parafoveal preview is also real. At moderate speeds, the trade is often favourable. The warpread.app RSVP reader lets you calibrate this precisely by controlling your WPM directly.

Regressions: the case against eliminating them

Some speed reading courses teach you to "train yourself not to go back." The eye movement research advises caution here.

Schotter et al. (2014) found that readers who were prevented from making regressions showed significantly worse comprehension than those who could re-read freely. Regressions are not a failure of discipline — they are a signal from your comprehension system that something upstream needs revisiting.

That said, undirected, habitual regressions — returning to words not because you failed to understand them but out of anxiety or distraction — do slow reading without benefit. The solution is not suppressing regressions by force of will but addressing the underlying cause: distraction, insufficient vocabulary for the domain, or reading at too high a speed for the material.

Applying this knowledge

Understanding the eye movement science gives you concrete levers:

• For difficult material (philosophy, law, dense non-fiction): accept longer fixation durations. Your reading speed on this material will be — and should be — lower than on narrative fiction. See our guide on reading speed for dense material.
• For narrative fiction: word skipping and parafoveal processing mean you can read faster with less comprehension cost. RSVP at 350–450 WPM works well for plot-driven texts. Try it on warpread with a classic like Crime and Punishment.
• For building speed: domain familiarity and vocabulary growth reduce fixation duration more reliably than any technique. Reading widely within a domain is the most evidence-backed route to reading that domain faster.

You cannot consciously control your saccades. But you can control the inputs — vocabulary, familiarity, WPM setting — that determine how efficiently your eyes move. That is the honest lesson of 130 years of eye movement research.

Try warpread.app free — RSVP reading at your own pace

References

• Rayner, K. (1998). Eye movements in reading and information processing: 20 years of research. Psychological Bulletin, 124(3), 372–422.
• Rayner, K., Schotter, E.R., Masson, M.E.J., Potter, M.C., & Treiman, R. (2016). So Much to Read, So Little Time. Psychological Science in the Public Interest, 17(1), 4–34.
• Schotter, E.R., Tran, R., & Rayner, K. (2014). Don't Believe What You Read (Only Once): Comprehension Is Supported by Regressions During Reading. Psychological Science, 25(6), 1218–1226.
• Clifton, C., Ferreira, F., Henderson, J.M., Inhoff, A.W., Liversedge, S.P., Reichle, E.D., & Schotter, E.R. (2016). Eye movements in reading and information processing: Keith Rayner's 40 years of research. Journal of Memory and Language, 86, 1–19.