Why Rereading Is Almost Always Less Effective Than Recall Testing

The testing effect is not new. Edwina Abbott in 1909, working under the influential educational psychologist Charles Judd, reported that students who practiced recitation after reading retained more than students who spent the same time reading. Gates (1917) found similar results with schoolchildren: the optimal ratio of study-to-recitation time tilted heavily toward recitation, not reading. Spitzer (1939) demonstrated the benefits of testing as a memory tool in a large-scale study of Iowa schoolchildren.

These findings were largely ignored. They didn't fit the dominant models of learning at the time, and the research didn't have the methodological sophistication to rule out alternative explanations convincingly.

Roediger and Karpicke's work starting in 2006 replaunched the question with cleaner experimental design and more compelling data. Their key insight was to use delayed testing — not just measuring what students remembered immediately after studying but what they remembered days and weeks later. Immediate tests favor rereading (or at least narrow the gap). Delayed tests reveal the substantial superiority of retrieval practice. This matters because "learning" that doesn't persist is not learning.

Subsequent researchers — Pooja Agarwal, John Dunlosky, Mary Pyc, Katherine Rawson, and others — have extended the findings across domains (medical education, language learning, science, history), formats (multiple choice, short answer, free recall), and populations (children, undergraduates, medical students, older adults). The effect is consistent enough that Dunlosky's influential 2013 review rated retrieval practice as "high utility" — one of only two strategies given that rating out of ten reviewed. The other was distributed practice (spacing).

To understand why rereading is so persistently popular despite its mediocre results, you have to understand the fluency illusion.

Fluency is ease of cognitive processing. Material that has been encountered before is processed more quickly and smoothly than novel material. This fluency generates a sense of familiarity that is easily confused with understanding or memory. When you reread a text and it processes smoothly, your metacognitive system interprets the smoothness as evidence of learning. You feel like you know it better.

This is a metacognitive failure — a failure to accurately assess your own knowledge state. The experience of the material becoming easier to process is real, but it reflects processing fluency, not durable memory. The knowledge encoded during rereading tends to fade quickly because it wasn't retrieved — it was just recognized.

Cognitive scientist Nate Kornell and colleagues demonstrated this in studies where students predicted their own memory after different study strategies. Students consistently predicted that rereading would help them remember as much as retrieval practice. After testing, their predictions were badly miscalibrated — retrieval practice substantially outperformed rereading, and students didn't see it coming.

The implication: your subjective experience of studying is an unreliable guide to how well you're learning. You need external feedback — actual retrieval tests — to calibrate your sense of what you know.

Several mechanisms have been proposed for why testing enhances memory. They're not mutually exclusive.

The desirable difficulties framework (Bjork and Bjork) holds that conditions that make encoding more difficult — like retrieval practice, spacing, and interleaving — produce better long-term retention even though they produce slower apparent learning. The difficulty is "desirable" because it triggers deeper processing. Rereading is too easy; the mental work is too low to generate strong encoding.

Elaborative retrieval: When you retrieve information from memory, you don't simply pull it out unchanged. The retrieval process involves reconstruction — connecting the retrieved information to surrounding knowledge, context, and schema. This reconstruction elaborates the memory trace and integrates it more deeply with existing knowledge. Rereading, by contrast, provides the information rather than requiring you to generate it, bypassing this elaboration.

Memory trace strengthening: Neural memory traces are strengthened by use. Retrieval is use; rereading is not (or is much less so). The act of retrieval reactivates the trace and triggers reconsolidation — a process by which memories are restabilized, often in a strengthened form. Reread-only strategies don't trigger this.

Inhibition and updating: Some researchers (Bjork) suggest that retrieval involves active inhibition of competing memories, which strengthens the target memory relative to competitors. This would explain why retrieval improves discrimination between similar concepts — not just retention of individual items.

Testing as a diagnostic: Retrieval practice also works as a feedback mechanism. When you attempt recall and fail, you learn exactly what you don't know. This focused feedback is more valuable for subsequent study than rereading the whole text, because you can direct attention to gaps. Rereading doesn't reveal your gaps — it obscures them behind fluency.

Retrieval practice is most effective when combined with distributed practice (spacing). The spacing effect — learning is better when study is spread across multiple sessions rather than concentrated into one — is also one of the most replicated findings in learning science.

The interaction is powerful: spaced retrieval practice, where you attempt recall of material at increasing intervals after learning, is dramatically more effective than either spacing alone or retrieval practice without spacing.

This is the scientific foundation for spaced repetition systems (SRS) like Anki. The algorithm presents flashcards at increasing intervals based on retrieval success. Items you retrieve easily get shown less frequently; items you struggle with get shown more frequently. This optimizes the spacing for each item individually, targeting retrieval when memory has decayed enough to make the retrieval effortful but not impossible.

Effortful retrieval at the right spacing interval is the most efficient path to durable memory that cognitive science has found. It's also boring, uncomfortable, and requires sustained effort without the feeling of momentum that rereading provides. This gap between optimal strategy and popular strategy is significant.

Dunlosky's 2013 review ranked ten commonly used learning strategies on utility. The results:

High utility: - Practice testing (retrieval practice) - Distributed practice (spacing)

Moderate utility: - Interleaved practice - Elaborative interrogation (asking "why" while reading) - Self-explanation (explaining material to yourself)

Low utility: - Summarization - Highlighting/underlining - Keyword mnemonics - Imagery use for text learning - Rereading

Highlighting and underlining made the low utility list, which may surprise people who annotate heavily. The problem: highlighting identifies that something is important without requiring you to do anything with it. It's a marker, not a learning event. Highlighted notes are often experienced as "done" — the material is marked, now it can be reread. They support rereading, not retrieval.

Self-explanation and elaborative interrogation made the moderate list. These require active engagement — you're generating something, not just consuming. The engagement is less than full retrieval practice but substantially more than passive rereading.

The blank page method is a practical implementation of free recall that works well for conceptual material:

1. Read a chapter or section. Take notes if useful. 2. Set the material aside. 3. Take a blank page. Without looking at anything, write down everything you remember about what you just read: main ideas, key claims, evidence, connections, implications. Don't organize it — just get it out. 4. When you've exhausted what you can retrieve, compare your blank page to the original. 5. Identify what's missing or wrong. Those are your actual study targets. 6. Reread only those specific gaps — not the whole material.

This protocol is superior to rereading because: it forces retrieval (which strengthens memory), it generates accurate metacognitive feedback (you can see exactly what you don't know), and it focuses subsequent study on actual gaps rather than re-exposing you to material you already know.

The blank page method also works at different scales: after a meeting (what do you remember from it?), after a conversation, after a book, after a course. It's a habit of mind more than a formal procedure.

One reason testing is underused: teachers don't want to make students feel dumb. Testing — especially in-class testing — creates anxiety, public failure, and discomfort. Rereading and lecture-with-notes feel supportive and structured.

The irony is that the discomfort of testing is the signal that learning is happening. The struggle to retrieve — feeling like you can't remember, reaching for something that isn't there yet — is the cognitive state that produces durable learning. Eliminating that struggle by providing the information (rereading, lecture) eliminates the learning mechanism.

Agarwal's work on retrieval-augmented learning in classrooms shows that low-stakes frequent testing dramatically improves retention without increasing anxiety (when it's framed as practice rather than evaluation). The issue isn't testing per se — it's high-stakes testing with social consequences. Low-stakes self-testing, which has no negative consequences, is purely beneficial.

For self-directed learners, this is controllable. You don't need anyone's permission to test yourself. The protocol is: read less, retrieve more. Every reading session should end with a retrieval attempt. Every review session should start with retrieval before rereading. The information is almost never as well-retained as it feels like during reading — and the only way to find out is to close the book and try.

The testing effect is partly a story about learning strategies. But it's also a story about self-knowledge. The people who keep rereading despite evidence that it doesn't work are making the same error repeatedly because their feedback loop is broken. They feel like they learned, so they believe they learned. The feeling is misleading them.

Robust self-knowledge about learning requires:

1. Recognizing that subjective experience of ease is not a reliable indicator of learning depth. 2. Using actual retrieval tests as the measurement tool, not subjective sense of familiarity. 3. Updating study strategies based on what the retrieval tests show, not what feels most comfortable.

This is metacognitive regulation — thinking about how you're thinking and learning, and adjusting based on accurate feedback. It's the same skill that underlies calibrated reasoning, evidence-based decision-making, and genuine expertise. The testing effect is just its application to the specific domain of memory and learning.

If you take one thing from this: the next time you're studying something important, close the material, take a blank page, and write down everything you remember. What you can retrieve is what you actually know. Everything else is an illusion of fluency.