Remembering something that never happened
Imagine a friend describing a wedding to you: bride, groom, cake, dancing, music… A few days later you may be sure they also mentioned the 'registrar', even though they never did. Memory does not work like a camera keeping a record of events; it stores meaning and fills the gaps with the most likely pieces. Most of the time that serves us well; sometimes it keeps a detail that never happened alive as a vivid memory.
In this experiment you read four short word lists. Each list is made up of associates of a word that never appears on it. In the test that follows you say whether words were on the lists and how sure you are. This is one of the most widely used methods for studying false memories in psychology: Deese–Roediger–McDermott, or DRM for short.
From Deese to Roediger and McDermott
In 1959 James Deese noticed something curious in free recall after word lists: after certain lists, people often wrote down a particular word that had not been on the list. The finding attracted little attention for a long time. In 1995 Henry Roediger and Kathleen McDermott took it up again and turned it into one of the basic methods of false memory research.
In their first experiment, participants heard six 12-word lists ('bed, rest, awake, tired, dream…' but no 'sleep') and wrote down what they remembered after each one. The hidden words were 'recalled' 40% of the time, about as often as real words from the middle of the lists. On the later recognition test, people said 'there' to 86% of the words they had heard and 84% of the hidden words, but to only 2% of words unrelated to the lists. More strikingly, for 58% of the hidden words people said they were sure.
Do you remember, or do you know?
In their second experiment Roediger and McDermott used longer, 15-word lists, and free recall of the hidden words rose to 55%. On the recognition test they asked participants to distinguish two kinds of experience for words they called 'old': 'remember' if they could mentally relive the moment the word was spoken, 'know' if they simply knew it had been there.
The result showed that false memories are not just guesses: people often said 'remember' for words they had never heard, reporting that they relived the moment they heard them. The authors summed up their paper with the line: people remember events that never happened. We do not ask 'remember or know' in this experiment, but the four-point confidence rating shows something similar: how often you say 'definitely there' to a hidden word.
Not every list is the same
In 1999 Michael Stadler, Roediger and McDermott published norms for 36 lists. All of them had been built to produce false memories, but their effects varied enormously: the share of people who freely 'recalled' the hidden word was 10% for the weakest list and 65% for the strongest, and false recognition ranged from 27% to 84%. The strongest was the 'window' list: after door, glass, pane, sill… 84% of people later said they had heard 'window' too. For people playing in English we chose lists from these norms.
In Turkish, in 2020 Meliscan Akdoğan, Ümit Akırmak and İdil Gürsoy built ten lists from Turkish association norms (Tekcan & Göz 2005). On their recognition test people said 'there' to 83% of studied words, 71% of hidden words and 6% of unrelated words. Again the lists differed greatly in strength: 'sleep', 'shirt', 'flag', 'cold' and 'window' produced the most errors, 'police' the fewest. For people playing in Turkish we use the five strongest of these lists and the 'car' list.
Why do we fall for it?
Two explanations are often given. In the first, every word on the list spontaneously activates the hidden word in the network of meanings; 'sleep' may well have crossed your mind as you read. Later, on the test, you cannot tell the source of that inner activation apart: did you see the word, or only think it? In the second, memory stores the gist of a list as well as its details: 'a list about sleep'. As the details fade, every word that fits the gist feels familiar.
Reviewing fifteen years of DRM research in 2010, David Gallo stressed that the illusion arises not from a single process but from several, sometimes opposing ones. He also reported that individual differences in susceptibility to this laboratory illusion have been linked to everyday errors such as misremembering public events. One important point: the DRM illusion is not a sign of carelessness or weakness. It is a by-product of a healthy memory that stores meaning and gist and can generalise.
What your score does and does not tell you
Your score is the number of the eight studied words you recognised, minus the number of the eight unstudied words (four hidden, four unrelated) you said were there; it runs from −8 to 8. It weighs hits against false alarms and sums up your result at a glance; it is not meant for rankings or races. The real result of the experiment is how often you said 'there' to the hidden words and how confident you were: falling for the hidden words while hardly ever falling for the unrelated ones shows that the error is not random but comes from meaning.
When you compare your result with laboratory values, keep the differences in mind: in the original experiments there was free recall after each list, which can increase false recognition, while we go straight to the recognition test. The words are written rather than spoken, and the lists are shorter. Because people who know the experiment will look out for the hidden word, we ask about this after the test and show them separately in the crowd. Your first game is the one kept in the scientific data; if you play again, you will know the trick and the result will be different.