Sophie

Research · Learning

Why we put writing first.

Writing is how you find out what you think. Here is the science behind it, and why we built Sophie around a sheet of paper.

A teenager on a window seat sketches a parabola in her notebook while Sophie rests beside her and rain runs down the glass.

Explaining a step on paper forces you to actually know it. Reading someone else's answer does not.

At Revyze, that idea shaped how we built Sophie, our AI math tutor. Sophie is writing-first: students work every problem by hand, on paper, and Sophie reads the page. We chose that on purpose.

Here is the science behind it, where it is strong, where it is thinner than the headlines, and why we made the call.

Writing is how you think

When you write something yourself, you remember it better and understand it more deeply.

Psychologists call the first part the generation effect. In 1978, Norman Slamecka and Peter Graf showed that people remember words they produce themselves better than words they only read. Producing takes effort, and the effort is what makes it stick.

The second part shows up in classrooms. In 2020, Steve Graham and his colleagues pooled school experiments in which students wrote about what they were learning. Writing about the content reliably improved learning, in math as much as in science and social studies.

So writing is more than a way to show what you know. It is one of the ways you come to know it. The next question is whether the hand matters, or only the words.

What the brain scans show

Only freehand writing switched on the reading parts of young children's brains. Typing and tracing did not.

At Indiana University, Karin James and Laura Engelhardt worked with 15 four- and five-year-olds who could not read yet. Each child practiced letters three ways: printing them freehand, typing them, and tracing over them. Then the children looked at letters inside an MRI scanner.

After freehand printing, the children used their brain's reading circuit to look at letters. These are the same regions skilled readers use. After typing or tracing, that circuit stayed quiet.

The tracing result is the surprising one. Tracing looks like writing, and it produces neater letters. But a child's own letters come out wobbly and a little different each time. James's lab later tested why that helps: seeing many imperfect versions of a letter teaches the brain what makes a "b" a "b." Perfect copies teach less.

The effect does not end in kindergarten. In 2024, Norwegian researchers Ruud van der Weel and Audrey van der Meer recorded the brain activity of 36 university students with 256 sensors. The students wrote words by hand, then typed the same words. Handwriting lit up wide connections across brain regions tied to memory and attention. Typing barely did.

Look at the tool they used: a digital pen on a touchscreen, not paper. The brain cared about the hand forming the word, not the surface. The authors' advice: "Children, from an early age, must be exposed to handwriting activities in school to establish the neuronal connectivity patterns that provide the brain with optimal conditions for learning."

Why a pencil or stylus, not a finger

A finger is a fine first tool for a four-year-old. After that, the hand needs something to hold.

One honest wrinkle first. In an eight-week study by Melissa Patchan and Cynthia Puranik, preschoolers practiced capital letters three ways: pencil on paper, stylus on a tablet, or fingertip on a tablet. The finger group wrote the most letters correctly (summary here). The researchers credit the feel of skin on glass, and the fact that a finger is easier to steer at that age. The stylus group matched the pencil group.

So a finger can help a small child meet letters. It is not where writing is going.

  • The grip trains the hand. Holding a pencil between thumb and two fingers builds the small muscles of the hand. That control is linked to school success later. Laura Dinehart and Louis Manfra tracked about 3,000 Miami children. Preschoolers with strong fine-motor writing skills became second graders averaging a B in math and reading. Those with weak skills averaged a C. Dinehart is careful to say this is a link, not proof of cause.
  • A fingertip is blunt. It is wide and it hides what it draws. It works for a big capital A. It cannot write a small "e", an exponent, or a fraction.
  • Most finger apps are tracing apps. Follow the dotted line, get a star. Tracing is exactly the practice that left the reading circuit quiet in the Indiana study.

The bottom line: pencil and paper is the gold standard. An iPad with a stylus is a good substitute, and the adult brain study above used one. A finger on glass is a starting point, not a school tool.

Slower writing, more thinking

Handwriting is slower than typing, and that is part of why it helps.

Nobody can write down everything a teacher says by hand. So a student has to choose, shorten, and put it in their own words. That squeezing is thinking. A keyboard lets you copy word-for-word without much of it.

That was the idea behind a famous 2014 study, "The Pen Is Mightier Than the Keyboard," by Pam Mueller and Daniel Oppenheimer. Students who took notes by hand beat laptop note-takers on questions about ideas. The laptop notes were more word-for-word.

Now the candid part. When Kayla Morehead, John Dunlosky and Katherine Rawson repeated the study in 2019, the gap mostly disappeared. It vanished on tests taken days later, and when students could review their notes. What mattered most was the quality of the notes.

So the honest claim is smaller than the headlines. A pencil does not make anyone smarter. It makes the lazy shortcut harder, and pushes you toward the choosing and summarizing that good notes need. The Norwegian researchers put it well: what matters is knowing "when to write by hand or use a digital device."

Why it matters even more in math

Math is where handwriting wins by the widest margin.

An equation is not a line of text. Fractions stack. Exponents float. A limit sits under its symbol. A keyboard flattens all of that into one line, and every symbol costs a menu click.

At Carnegie Mellon, Lisa Anthony, Jie Yang and Ken Koedinger timed people entering calculus-level equations. Writing them by hand was about three times faster than typing them into an equation editor. The harder the equation, the bigger the gap. In a later algebra study, students who worked by hand learned as much as students who typed, in about half the time.

Paper also holds what a keyboard can't: a quick sketch of a graph, an arrow from one line to the next, a mistake crossed out but still visible. That crossed-out line shows a teacher exactly where the thinking slipped.

What parents can do

You don't need to ban screens. You need to keep a pencil in the loop.

  • Keep paper and pencils where homework happens. If they are in reach, kids use them.
  • Ask for freehand letters, not only tracing. Tracing sheets are fine for a warm-up. The learning happens when the letter comes out of the child's own hand.
  • If the work lives on a tablet, buy the stylus. A stylus on glass counts as handwriting. A finger mostly doesn't, past preschool.
  • Type to publish, write to learn. A finished essay can be typed. Notes, outlines and practice problems go by hand.
  • For math, always show work on paper. Every step, including the wrong ones.

Why Revyze is writing-first

We could have built Sophie as a chat box: type a question, get an answer. We built her around a sheet of paper instead, for four reasons.

  • The exam is on paper. AP Calculus is a hybrid exam. Students read the free-response questions in an app, then handwrite their answers in paper booklets. That section is half the score. Practice should look like the test.
  • The work is the lesson. A typed final answer tells a tutor almost nothing. A page of working shows the setup, the step that slipped and the line crossed out. Sophie reads the page and points at the slip, not just the result.
  • The student does the producing. Tools that answer on demand drift toward handing out answers. When the student writes every step, they do the generating, and that is the part that sticks.
  • It runs on what families already own. A pencil, a sheet of paper, and the family's phone on a stand as Sophie's eye on the desk. No special tablet needed.

Screens are good at a lot of things. Working through a calculus problem one line at a time is still the pencil's job.

Sources