Memory & Attention · Quick Test
Generation Effect
Why does a word you complete yourself stick better in memory?
The experiment
Twelve pairs of words will appear, one at a time, at your own pace. In some, the second word comes ready: just read it. In others, you write the opposite of the first word, helped by the first letter. Try to keep the second words in mind.
What happened?
On average, people remember the words they completed themselves better than the ones they only read, even when completing them was easy and the word was almost obvious. In a round with only twelve words, chance weighs a lot and the difference may not even show up. Across thousands of people, the advantage of generating is one of the most consistent findings in memory research.
Why does your brain do this?
To write the opposite of “tall”, the mind has to understand the word, search memory for what fits the rule and check that the first letter matches. That work involves the meaning of the word and links it to the cue. Reading the finished pair asks for much less: the word passes before your eyes without you having to look for it. There are several explanations for what exactly that effort changes, and researchers still debate which one matters most.
When does this shortcut work well?
Giving more weight to what took effort makes sense: in general, what you produced, decided or discovered yourself is more likely to matter to you than something that just passed before your eyes. The effect is an advantage, not a flaw, and you can use it in your favor.
Where does this show up in daily life?
Rereading your notes feels like studying, but trying to answer from memory, filling in a summary with gaps or explaining the topic in your own words usually leaves more in memory. The same goes for writing down a route instead of just following the GPS. When you want to remember something, turn the reading into a question.
Deep Dive
Open Nerd Mode
The generation effect got its name from Norman Slamecka and Peter Graf in 1978, and was also described the same year by Larry Jacoby. In the classic format, people receive pairs of words: half come complete, to be read, and in the other half they produce the second word from a rule (a synonym, an opposite, a rhyme) and a cue. Then a test measures how much they remember of each kind.
Sharon Bertsch and colleagues (2007) pooled 86 studies, with 445 measures of the effect and 17,711 participants. On average, 55.9% of generated words were remembered, against 47.1% of read ones: an advantage of almost half a standard deviation (d = 0.40). The effect shows up in recognition tests, in cued recall and in free recall, which is the case in this experiment: in 215 measures with 9,091 people, 40% of generated words against 33% of read ones.
The meta-analysis also showed that the size varies a lot with the study design. The effect is larger when the same person reads some words and generates others, as here (d = 0.50), than when one group only reads and another only generates (d = 0.28). It was also larger with numbers than with words and nearly vanished with nonwords (d = 0.05), which have no meaning to search for in memory.
Technical terms
- Memory
- Recall
- Encoding
Discovery
Described by Norman Slamecka e Peter Graf in 1978.
Level of evidence
Robust evidence
Replicated in multi-lab studies or consistent meta-analyses.
Basis: Bertsch, S.; Pesta, B. J.; Wiscott, R.; McDaniel, M. A. (2007). The generation effect: A meta-analytic review.
References
- Original studySlamecka, N. J.; Graf, P. (1978). The generation effect: Delineation of a phenomenon. Journal of Experimental Psychology: Human Learning and Memory, 4(6), 592–604.
- ReplicationBertsch, S.; Pesta, B. J.; Wiscott, R.; McDaniel, M. A. (2007). The generation effect: A meta-analytic review. Memory & Cognition, 35(2), 201–210.
Review
This text has not been reviewed by a volunteer yet. Want to help review it?
Found a mistake? Write to [email protected]