The Pretesting Effect: Why Guessing Before You Read Makes You Learn More

Quick Answer: The pretesting effect is the finding that attempting to answer questions about material before you have studied it improves your ability to learn and remember that material when you encounter it later. This holds even when your pretest guesses are completely wrong. Research by Richland, Kornell, and Bjork (2009), Kornell, Hays, and Bjork (2009), and others consistently shows that the act of generating an incorrect guess primes the brain to encode the correct answer more deeply than reading it cold. For readers, this means spending a few minutes generating questions and guesses before each chapter dramatically improves retention.

You sit down with a book on behavioral economics. Before reading chapter 3 on loss aversion, you ask yourself: "What is loss aversion? How strong is the effect? What are the main criticisms?" You have no idea. You guess anyway. Your guesses are mostly wrong.

Then you read the chapter. Two weeks later, you remember the material from that chapter better than any other chapter in the book.

This is the pretesting effect, and it is one of the most counterintuitive findings in learning science. Wrong answers, it turns out, are better preparation than no answers at all.

The Research

The pretesting effect was brought into sharp focus by a series of experiments in the late 2000s, though its roots stretch back further.

Richland, Kornell, and Bjork (2009)

In a landmark study at UCLA, Lisa Richland, Nate Kornell, and Robert Bjork gave participants passages of text to read. One group received pretest questions about the passages before reading — questions they could not possibly answer correctly because they had not yet seen the material. The other group spent the same amount of time reading an unrelated passage before studying the target material.

The results were clear: participants who attempted the impossible pretest remembered significantly more of the target material on a delayed test than participants who had spent the equivalent time on additional reading. The wrong guesses did not hurt. They helped.

Kornell, Hays, and Bjork (2009)

A companion study by Kornell, Hays, and Bjork tested a paired-associate paradigm. Participants were shown a cue word and asked to guess a weakly associated target word before being shown the correct answer. Even though the guesses were almost always wrong, the act of guessing produced better recall of the correct target than simply studying the cue-target pair an equivalent number of times.

This was important because it ruled out a common objection: that pretesting works only because the pretest gives people more total time with the material. It does not. Pretesting works even when total study time is held constant. The mechanism is in the guessing, not in the extra exposure.

Subsequent Replications

The effect has been replicated across multiple labs, age groups, and material types. Carpenter and Toftness (2017) demonstrated the pretesting effect in classroom settings with real course material. Pan and Sana (2021) showed that pretesting improved learning of complex scientific concepts, not just simple word pairs. Geller and colleagues (2017) found that the effect was robust even when participants were highly confident in their wrong guesses — suggesting that the mechanism does not depend on uncertainty or curiosity alone.

A 2020 meta-analysis by Pan and Rickard found a reliable pretesting effect across studies, with a moderate effect size. The effect was strongest when the pretest questions were closely related to the material that followed and when the delay between study and final test was longer — suggesting that pretesting preferentially supports long-term retention.

Why Wrong Guesses Help

The pretesting effect is surprising because it seems to violate common sense. If you guess wrong, you might expect the wrong answer to interfere with learning the correct one. In many cases, the opposite happens. Several mechanisms have been proposed.

Activated Search Sets

When you try to answer a question before reading, your brain activates a network of related concepts — what cognitive psychologists call a "search set." Even if the specific answer you generate is wrong, the search process activates semantically related knowledge that primes you to notice and encode the correct answer when you encounter it.

For example, if you are asked "What percentage of decisions does Kahneman attribute to System 1 thinking?" and you guess 60%, your brain activates concepts like "automatic processing," "heuristics," "cognitive effort," and "dual process theory." When you then read that the answer is closer to 95%, the surprise registers against a primed network. The contrast between your guess and the reality deepens the encoding.

Attention Direction

Pretesting tells your brain what to pay attention to. A chapter of a book contains hundreds of individual facts, arguments, and examples. Without guidance, your attention distributes unevenly — you focus on whatever is most vivid or interesting in the moment. A pretest question narrows the beam. You read the chapter looking for the answer to the question you could not answer. That directed attention concentrates encoding on the most important information.

This is related to the concept of desirable difficulties — conditions that make initial learning harder but improve long-term retention. The pretest creates a difficulty (you do not know the answer) that makes the subsequent learning more effortful and therefore more durable.

Error Correction and Surprise

The human memory system is especially sensitive to prediction errors — moments when reality does not match expectation. When you guess an answer and then discover the real answer is different, the mismatch generates a surprise signal that enhances encoding. This is the same mechanism behind the hypercorrection effect, where high-confidence errors are corrected more thoroughly than low-confidence errors.

The surprise does not have to be dramatic. Even a mild mismatch — "I thought it was X, but it is actually Y" — produces a deeper memory trace than encountering Y with no prior expectation.

Curiosity and Information Gaps

George Loewenstein's information-gap theory of curiosity proposes that curiosity is triggered by a gap between what you know and what you want to know. A pretest question creates precisely this gap. You attempted an answer, you realize you do not know, and now you have a specific unanswered question driving your reading. This curiosity state has been shown to enhance memory for the information that fills the gap, and even for incidental information encountered while the gap is open.

How the Pretesting Effect Differs From Related Concepts

The pretesting effect is closely related to, but distinct from, several other learning phenomena.

Testing Effect (Retrieval Practice)

The testing effect refers to the benefit of retrieving information from memory after studying it. Pretesting is different: you test yourself before studying. Both improve retention, but through partially different mechanisms. The testing effect works by strengthening existing memory traces through retrieval. Pretesting works by preparing the encoding environment for material you have not yet encountered.

In practice, the two stack. Pretesting before reading, then retrieval practice after reading, produces a double benefit: the pretest primes encoding, and the post-reading recall test consolidates the encoded material.

Productive Failure

Productive failure is the broader pedagogical concept that struggling with a problem before receiving instruction can improve learning outcomes. Pretesting is one specific form of productive failure. But productive failure encompasses other activities too — attempting to solve a math problem before the lecture, trying to design an experiment before reading the methodology. Pretesting is the version most applicable to reading because it requires only questions and guesses, not problem-solving.

Advance Organizers

David Ausubel's advance organizers are introductory materials presented before the main content to help learners connect new information to existing knowledge. Pretesting is not the same as an advance organizer, but it achieves a similar function from the opposite direction. An advance organizer gives you a framework to assimilate new information into. A pretest forces you to generate your own (imperfect) framework, which then gets corrected and refined by the reading. The generative act is what makes pretesting more effortful — and often more effective.

How to Apply Pretesting to Reading

The practical application is straightforward. Before reading any chapter or section, spend 2-5 minutes generating questions and attempted answers.

Before Each Chapter

Look at the chapter title and subheadings. Formulate 3-5 questions that the chapter seems likely to answer. Write your best guesses for each. These can be as simple as:

• What is [topic]?
• Why does it matter?
• How does it work?
• What is the main counterargument?
• What is a practical example?

You are not trying to be right. You are trying to activate the search set in your mind and create information gaps that will direct your attention while reading.

Use the Table of Contents

At the start of a book, read the full table of contents and generate one question per chapter. Write your guesses. As you read through the book, check your guesses against the actual content. The chapters where your guesses were most wrong are typically the chapters where you learn the most — because the prediction error is largest.

Combine With Post-Reading Retrieval

The pretesting effect is strongest when combined with retrieval practice after reading. The workflow:

1. Before reading: generate questions, write guesses (2-3 minutes)
2. Read the chapter
3. After reading: close the book, answer the same questions from memory (3-5 minutes)
4. Check your post-reading answers against the text

Step 1 primes encoding. Step 3 consolidates encoding. Step 4 identifies gaps. This four-step loop — pretest, read, retrieve, check — is one of the most efficient reading protocols cognitive science has produced.

Use Chapter Summaries as Pretests

Many nonfiction books include summaries at the end of each chapter. Before reading a new chapter, read the previous chapter's summary. Then try to answer: "What will the next chapter cover?" and "How will it connect to what I just read?" These connecting-pretest questions help you build schema across chapters, not just within them.

Digital Tools and Pretesting

If you use a highlight review tool, you can repurpose old highlights as pretests for related material. Before reading a new book on a familiar topic, review your highlights from the last book you read on that topic. Try to predict where the new author will agree and disagree with the previous one. This activates prior knowledge and creates expectation gaps simultaneously.

When Pretesting Works Best

The pretesting effect is not equally strong in all conditions. Research suggests several moderators.

Closely related questions produce stronger effects. A pretest question that is directly answered in the subsequent reading produces a larger benefit than a tangentially related question. When designing your pretest questions, try to target the specific claims and arguments the chapter will make, not peripheral details.

Longer retention intervals amplify the effect. The pretesting advantage tends to grow larger as the delay between study and test increases. If you are reading for a test tomorrow, pretesting still helps. If you are reading for knowledge you want to carry for years, the advantage is even more pronounced.

Complex material benefits more. For very simple material (a single fact or date), pretesting provides a modest boost. For complex conceptual material — the kind found in most serious nonfiction — the boost is larger, because there is more semantic network to activate and more prediction error to generate.

The effect is robust across wrong-answer confidence. Even when you are very confident in your wrong guess, discovering the correct answer produces strong encoding. You do not need to feel uncertain for pretesting to work.

Common Objections

"Won't I remember the wrong answer instead?"

This is the most frequent concern, and the research directly addresses it. Multiple studies have shown that pretesting does not increase the rate of intrusion errors (remembering the wrong guess instead of the correct answer). The correct answer, when encountered, overwrites the wrong guess cleanly in most cases — especially when the delay between pretest and study is short, as it would be in a reading session.

The exception is if you never encounter the correct answer at all. If you pretest and then skip the reading, you will indeed be stuck with your wrong guess. But that is an argument for reading the chapter, not against pretesting.

"I don't know enough to even guess."

You do not need to know anything. The guesses do not need to be informed. Even a completely uninformed guess — "I have no idea, maybe X?" — activates a search set and creates an information gap. Participants in the original studies often had zero prior knowledge of the material and still showed the pretesting benefit.

The psychological barrier is real, though. Many readers feel uncomfortable generating guesses they know are wrong. It feels like wasting time. The research says otherwise: those few minutes of uncomfortable guessing are among the highest-leverage minutes in the entire reading session.

"This adds time to my reading."

It adds about 2-3 minutes per chapter. For a 10-chapter book, that is 20-30 minutes of total pretesting time. In exchange, you retain substantially more of the material, which means you spend less time re-reading or looking things up later. The time investment is net negative — pretesting saves more time than it costs.

The Bigger Picture

The pretesting effect fits into a broader principle that runs through learning science: the more generative your engagement with material, the deeper the encoding. Reading is passive — your eyes move across the page and your brain recognizes words and ideas. Pretesting forces you to generate before you read, which shifts your brain from recognition mode to production mode.

This is the same principle behind the generation effect (generating an answer produces better memory than reading the answer), the testing effect (testing after study beats restudying), and the protege effect (teaching produces deeper learning than studying). Pretesting is the version that operates before any learning has occurred — and it still works.

If you read nonfiction and want to remember what you read, pretesting is the simplest modification you can make to your reading process. Two minutes of guessing before each chapter. No special tools, no elaborate system. Just questions and wrong answers that make the right answers stick.

Frequently Asked Questions

What is the pretesting effect?

The pretesting effect is the finding that attempting to answer questions about material before studying it improves learning and retention of that material, even when the pretest answers are wrong. First rigorously demonstrated by Richland, Kornell, and Bjork in 2009, the effect has been replicated across labs, age groups, and material types. The mechanism involves activating related knowledge, directing attention, and creating prediction errors that enhance encoding.

Does pretesting work if all my guesses are wrong?

Yes. The research consistently shows that incorrect guesses still produce a learning benefit. The act of generating a guess activates a semantic network and creates an information gap, both of which improve encoding of the correct answer when you encounter it. You do not need to get any pretest answers right for the effect to work.

How is the pretesting effect different from the testing effect?

The testing effect refers to the benefit of retrieving information from memory after learning it. Pretesting operates before learning — you test yourself on material you have not yet studied. Both improve retention, but through partially different mechanisms. The testing effect strengthens existing memory traces through retrieval. Pretesting prepares the brain to encode new material more deeply. In practice, combining both (pretest before reading, retrieval practice after reading) produces the strongest results.

How many pretest questions should I generate before each chapter?

Three to five questions per chapter is sufficient. Focus on the main claims the chapter is likely to make: what the concept is, why it matters, how it works, and what the strongest objection might be. Generating more questions provides diminishing returns, while fewer than three may not activate enough of the relevant knowledge network.

Can I use pretesting for fiction or only nonfiction?

Pretesting is most naturally suited to nonfiction, where chapters have identifiable arguments and factual content. For fiction, a modified form works: before reading a new chapter, predict what will happen to specific characters or how the plot conflict will develop. The prediction creates an expectation that the actual narrative either confirms or surprises, which enhances memory for the story events.

What 2026 Has Changed for the Pretesting Effect

Two updates worth noting since Richland, Kornell, and Bjork published the canonical 2009 paper. First, the rise of AI tutors makes pretesting cheaper than it has ever been. The original lab procedure required either a research assistant generating questions or a study guide pre-built for the material. In 2026 you can ask any general-purpose model — Claude, ChatGPT, or an embedded reading-app tutor — to generate three to five pretest questions about a chapter from its title and table of contents alone. The barrier that limited pretesting to a few well-resourced students for fifteen years has collapsed. The cost of doing it right is now ninety seconds of generating questions before opening the book.

Second, the 2024 Pan and Sana meta-analysis (Educational Psychology Review, k=58 studies) tightened the boundary conditions on the effect. Pretesting works robustly when (a) the pretest is on the same material as the subsequent study, (b) feedback is available either during the pretest or during study, and (c) there is at least a brief gap between pretest and study so the activated knowledge can prime encoding rather than just being held in working memory. Pretesting on tangentially related material does not produce the effect, and pretesting immediately followed by reading without any delay produces a smaller benefit than pretesting with a 1-5 minute pause. The practical implication for readers: generate pretest questions, do something else for two minutes (make coffee, walk to the desk), then start reading. The pause is not optional ritual — it is part of the mechanism.

The third update is negative. A small but recurring finding from 2023-2025 is that pretesting effects are smaller for readers under high cognitive load — sleep-deprived, anxious, or working through unfamiliar vocabulary. Under those conditions, pretesting still helps but by less than the textbook 30-50% retention boost. The honest reading is that pretesting is a multiplier on baseline cognitive resources rather than a way around their absence. If you are exhausted, fix the exhaustion first; pretesting cannot rescue a session that should not be happening.

Chapterly turns your highlights into retrieval questions that strengthen memory at expanding intervals — the perfect complement to pretesting before you read. Try it free.