How does "Prioritize items you were confidently wrong about" work?

The hypercorrection effect is the specific finding that high-confidence errors are better corrected and retained than low-confidence errors or uncertain guesses. The mechanism is a combination of surprise (high confidence + failure = strong prediction error) and attention: when you are shocked to be wrong, you attend more to the correction. Butterfield and Mangels documented that high-confidence errors with feedback show especially strong subsequent retention.

How does "Generate your own examples or explanations before studying provided ones" work?

The generation effect (Slamecka & Graf, 1978) shows that self-generated material is better retained than received material, independent of the error dimension. Combining generation with the errorful learning effect — generating your own example, then comparing it to a better one — produces encoding from multiple sources: generation effort, prediction error, and contrast-based learning. The comparison step is where the structural principles become salient.

How does "Receive corrective feedback promptly after a test attempt" work?

The prediction error signal created by an error is strongest immediately after the error. Prompt feedback catches the brain while the error representation is still active and the "correction needed" signal is salient. Delayed feedback allows the error to consolidate partially, requiring more effort to overwrite. For very confident errors specifically, the hypercorrection mechanism depends on the emotional salience of being corrected while the surprise is still present.

How does "Reframe difficulty and errors as the mechanism, not the obstacle" work?

Robert Bjork’s "desirable difficulties" framework shows that conditions that make learning feel harder in the short term (spacing, interleaving, generation, testing) consistently produce better long-term retention. The subjective experience of difficulty is not a signal that something is wrong — it is the mechanism. This reframe matters because learners who interpret difficulty as a sign of inadequacy tend to abandon the harder condition for easier (but less effective) study methods.

How does "Know when errorless learning is the right call instead" work?

Errorless learning (carefully guiding production to avoid errors) was developed for populations — including those with amnesia, severe cognitive impairment, and some dementia presentations — where the implicit memory systems that benefit from errorful approaches are impaired. For these populations, errors cannot be effectively used because error monitoring and the prediction signal mechanism are compromised. For healthy adult learners of conceptual material, errorful learning is generally superior; for procedural motor skill acquisition or clinical populations, the picture is more mixed.

Nate Kornell

How does making errors during learning improve long-term retention?

Prediction errors, the hypercorrection effect, and how to design your learning around productive failure

Short answer

Errorful learning is the counterintuitive finding — supported by Nate Kornell and others — that generating incorrect answers before receiving the correct one produces better long-term retention than studying the correct answer directly. The error creates a memory "prediction error signal" that primes deeper encoding of the correction. This contradicts the instinct to protect learners from being wrong, but the effect is real and robust under specific conditions.

Errorless learning — carefully guiding students to only produce correct responses — became popular in some educational and therapeutic contexts because wrong answers seemed harmful. The memory research, particularly Kornell’s work, tells a more nuanced story: errors that are recognized and corrected can actually strengthen retention of the correct answer more than error-free presentation. The key is the error must be followed promptly by corrective feedback. The practices below make productive error generation a deliberate learning strategy.

The practices (6)

Why it works

A guess activates memory networks related to the question. When the correct answer then arrives, it is encoded against the background of the activated-and-failed guess, creating a stronger prediction error signal. This signal marks the correct answer as important to remember and drives deeper processing. Research by Kornell et al. showed that even when guesses were wrong 100% of the time, retention of the correct answer was better than in a study-only condition.

How to do it

1Before looking anything up or receiving any explanation, write or say your best answer to the question.
2Rate your confidence 0–100%.
3Then receive the correct answer.
4Note which items you were confidently wrong about — the hypercorrection effect makes these especially well retained.

Evidence

RCT / meta-analysis

Kornell, Hays & Bjork (2009) showed that generating errors followed by corrective feedback produced better retention than studying correct answers directly, even for general-knowledge items where guesses were rarely correct.

Honest caveat: The errorful learning benefit depends critically on receiving correct feedback promptly. Errors without correction can embed false memories rather than enhance true learning.

— Kornell, Hays & Bjork (2009), Unsuccessful retrieval attempts enhance subsequent learning, Journal of Experimental Psychology: Learning, Memory, and Cognition

Common mistake: Treating the guess as a commitment ("I was wrong, I’m bad at this") rather than as a deliberate encoding tool — the emotional response to being wrong can short-circuit the memory benefit.

Go deeper on this practice →

Practice this with IX Coach

Reading about a practice changes nothing on its own. IX Coach turns these into a guided, adaptive routine — discerning where you are in real time and walking the practice with you, session after session.

Start with IX Coach →