Daniel Hawes

Daniel R. Hawes Ph.D.

Quilted Science

Confidently Wrong?

Correcting false memories

Posted May 12, 2010

Voldemort was near, so Harry slid underneath his invisibility cloak...but let me get back to this later.

Memory is a funny thing. Many people remember seemingly random minutiae from their childhood, but then fail to remember things that are actually quite important to them. On the other hand we're all also fairly prone to—every now and then—vividly remember things that never really happened.

For example, it has been shown that questions such as "How fast was the sports car passing by the barn while traveling along the country road" later makes people report having seen a nonexistent barn in an earlier video presentation.

Or exposing people to words such as "bed," "tired" and "rest" (or similar word groups) leads people to believe that they also saw the word "sleep".

Yet another example would be if I asked you to fill in the blank in the following sentence (from memory):
Voldemort was near, so Harry _____ underneath his invisibility cloak.

The problem with false memories—as with incorrect knowledge—is that they are often maintained even in the light of contradicting evidence. So researchers Lisa Fazio and Elisabeth Marsh may have come across something of general interest in their recent study of what type of false memories are more likely to be discarded, once correcting evidence is provided.

Take for example the above sentence completion task. You may have gotten the correct answer ("slid"), simply a wrong answer (e.g. "knelt") , or a (wrong) answer that may be categorized as non-studied inference (e.g. "hid").
Calvin and HobbesAdditionally, you may have been more or less confident in the answer you gave. (For a more comprehensive treatment of being confident in having it all wrong, please see here)

According to Fazio and Marsh, it may be the interplay between the types of memory error we make and our confidence in the answers we provide that determine whether we maintain our falsely held memories.

For a study set to appear in the next issue of the journal Psychological Science, the researchers contemplated whether the well-studied concept of hypercorrection in the knowledge domain (e.g. people who falsely believe that Einstein had poor grades in high school, will—once corrected—better remember the truth that Einstein was an A-student, than people who held no previous notions about Einstein's grades), may also extend to the domain of memories. In doing so they hypothesized that—once exposed as being incorrect—

"confidently held false memories should be corrected more often than other errors."

To test this hypothesis 46 undergraduate students were asked to study 48 sentences such as "The karate champion hit the cinder block" and then asked to perform memory-based sentence completion tasks such as the one in my Potter example (Note: Judging from recent grading experiences this is an almost impossible task for today's average undergraduate student!).

Subjects were also asked to grade their confidence in each answer on a 7-point scale.

After filling out the blank and noting their confidence the original sentence was shown again for 4 seconds, and after all 48 sentences had been completed, subjects were retested on the same 48 sentences; the goal being to identify the contingencies under which students learned from their mistakes.

The researchers found that 51% of entries in the first trial of 48 sentences were non-studied inferences (e.g. "kick" in the karate champion sentence), and that only 25% of answers were correct. However, participants learned from their errors so that correct answers made up 74% of responses in the second round (14% inferences).

In looking more closely at which errors had been corrected, the researchres found that

"high-confidence memory errors were more likely to be corrected on the final test than were low-confidence memory errors."

"Following feedback, subjects corrected more false memories (made with high confidence) than erroneous guesses. The implication for other episodic memory errors is that corrections will be most likely when feedback contradicts subjects' expectations. Hypercorrection of false memories is consistent with the idea that people attend more to feedback when it is surprising."

The researchers also made sure to gather information to help rule out other serious explanations for their findings such as that

"the hypercorrection effect occurs because confidence in errors is correlated with knowledge about the target domain. For example, most readers of this [blog] will be more confident when answering questions about psychology than when answering questions about chemistry. However, they will remember feedback concerning an error in psychology better than feedback concerning an error in chemistry because the psychology feedback can be associated to their preexisting knowledge"

To account for this possibility, the study controlled for background knowledge, so that although

"[a]ll false memories depend on activation of meaning structures, [...] this is uncorrelated with confidence in episodic memories. Our finding of hypercorrection for episodic memories means that differences in domain knowledge cannot be solely responsible for the hypercorrection effect."

This is something worth remembering...I am certain!

Main Reference:

Lisa K. Fazio, & Elizabeth J. Marsh (2010). Correcting False Memories Psychological Science : 10.1177/0956797610371341

About the Authors

Kimberlee D’Ardenne, Ph.D.

Kimberlee D’Ardenne, Ph.D. is a neuroscientist by training, science writer by choice.

Rachael Grazioplene

Rachael Grazioplene is a graduate student in Psychology at the University of Minnesota. Her research is focused on individual differences and behavioral genetics.

Daniel Hawes

Daniel R. Hawes is a social psychologist stuck in an applied economist's body.

More Posts