Does Confidence in Your Memory Mean That You Are Correct?


The accuracy of an eyewitness’ memory is open to question. 

Confidence in one’s memory is generally believed to be a sign of memory accuracy, but this is not always the case. Some studies have shown a high correlation between confidence and accuracy, while others have shown little to no correlation.

New experimental research by Roediger & DeSoto (2014), published in the journal Memory, has examined the relationship between confidence and memory accuracy using a word recognition memory paradigm.

In short, the study shows that very confident people tend to be more inaccurate, compared to less confident people, when they distinguish between strongly related objects.

Now to the procedure of the study:

  • First, subjects were shown 150 words within 10 different categories, e.g a bird category. Some words were more dominant than others, e.g. eagle is more dominant than flamingo in a bird category
  • The subjects were then distracted for five minutes to make recall more difficult. After the distraction, they were presented with 300 words, some familiar (old) words and some new ones (lures). The lures were either unrelated, strongly or weakly related to the words that they were shown at first
  • Now the subjects were told to judge whether the words were new or old. After having made this old/new judgment, they were asked to rate their confidence in their judgement on a scale from 0  to 100

In sum, the research shows that people who are more confident in their memory also tend to be more accurate. However, this finding was only true for weakly related or unrelated lures.

Technically, the researchers found a negative correlation between confidence and accuracy for strongly related lures. For strongly related lures, subjects who were more confident were less accurate than subjects who were less confident. That is an interesting finding since it is counterintuitive. For weakly related lures, there was a modest positive correlation between confidence and accuracy.

In other words, when people are shown unrelated lures, compared to strongly related lures, they make more accurate judgments. This makes sense in that it is easier to distinguish objects that are weakly related to each other compared to objets that are strongly related. The authors emphasize that these findings have implications for eyewitness memory situations:

“… The greater the similarity between a person in a line-up and the actual suspect, the more likely (all else being equal) that a false identification will occur with high confidence.”

They continue:

“Although the point seems obvious, we can find few experiments that actually demonstrate it, perhaps because confidence ratings are not generally taken in eyewitness research.”

Photo: Isaac Leedom