Some insight into the nature of memory

When I started work on a memoir about my childhood, I thought remembering what actually happened would be the easy part. I had very specific memories of very specific events, and I wrote them down exactly as I remembered them. One memory involved my stepsister winning Glen Campbell's Rhinestone Cowboy album at a pumpkin carving contest in 1974. I remember it as clear as the day it happened. Only Rhinestone Cowboy was released in 1975.

At about that same time, Fergus Craik and Endel Tulving were conducting the experiment that formed the basis a new framework of human memory: we don't remember things by just "looking up" the memory in our mind, as if we were simply seeking an index card in a recipe box. Instead, we actively assemble memories from individual bits of information, jumbled together and loosely associated. Whether we remember one thing or another depends not so much on whether the index card got placed in the box, but the type of associations we form between the items in our memory.

What Craik and Tulving did is to ask people to study a list of words. To help recall each word, one of three different types of questions was asked: either a question about the letters in the word (are there 6 letters in "salmon"?), the sound of the word (does this word rhyme with "backgammon"?), or the meaning of the word (is a salmon a fish?). Participants were then tested on the words they had studied. Either a new word or a previously studied word was presented, and respondents had to indicate whether the word was new or old. Participants remembered 80 percent of the old words when they had been asked a question about its meaning, but only 20 percent of the old words when they had only been asked about the letters in the word. Craik and Tulving reasoned that people remember words better when they have some "deep" association with the word. This interpretation of the results was controversial, but it nonetheless changed the way psychologists thought about memory. While the components of memory all might be stored the same way, the way they are retrieved requires the active participation of the person doing the retrieving. We build memories based on how strongly we associate them with other things.

Antonia Kronlund and Bruce Whittlesea of Simon Fraser University admired Craik and Tulving's work, but noted that their model could not explain false memories such as my memory of my stepsister. After all, by definition the participants never had an opportunity to form a deep association with a new word—they had never seen it before. Kronlund and Whittlesea developed a new test that also allowed them to test for false memories ("Seeing Double: Levels of Processing Can Cause False Memory," Canadian Journal of Psychology, 2005).

Kronlund and Whittlesea asked participants the same deep or shallow questions about a list of words, but they also repeated half of the words, asking participants only to say the word out loud. For the memory test, people were asked if they had seen a word once or twice. As in the original research, respondents responded correctly more often when a word with a deep question had been presented twice. However, they also responded incorrectly more often for words associated with deep questions that had only been presented once. This table summarizes some of their results. The percentage figures indicate how often respondents said they had seen a word twice, so the top row of the table indicates when people were answering incorrectly, and the bottom row indicates correct responses.

So participants generate 20 percentage points more correct responses when they have been asked a deep question about the word, but they also generate 12 percentage points more incorrect responses. Almost all the advantages of the deep question are made up for with additional false memories.

Kronlund and Whittlesea suggest that the deep questions don't simply increase the likelihood of retaining a memory, they also make that memory "feel special." It's possible that I remember my stepsister winning the prize because it was a special moment for me. However, it's also possible that the very special feeling associated with that memory is what led me to incorrectly recall the year in which it occurred.