Can You Trust Yourself?

New studies show how easily our memories can go astray.

Posted Mar 10, 2015

michaeljung/Shutterstock
Source: michaeljung/Shutterstock

One of the scariest parts of the legal system is its reliance on eyewitness testimony.  A witness who identifies a defendant as the perpetrator of a crime can sway a jury, in the absence of any physical evidence, that the defendant was actually the one who committed the crime.

For several decades, of course, we have known that eyewitness memory is faulty.  Classic studies by Elizabeth Loftus and her colleagues in the 1970s demonstrated that people would mix together information they saw and things they heard in later questions when thinking back to an event. In a 1974 paper written with John Palmer, participants watched a film of a car accident. Later, they were asked to judge how fast the cars were going. Some people were asked how fast the cars were going when they hit each other, while others were how fast they were going when they smashed into each other. A week later, participants were asked whether the saw broken glass in the accident. Those who were asked about the cars smashing into each other were much more likely to say they saw broken glass than those who were asked about the cards hitting each other.

On the basis of results like this, there are two possibilities. One is that when we remember things, we recreate our memory based on fragments of actual memories from the past. This view of memory suggests that we may make mistakes when we do this reconstruction, but somehow the truth is still buried in our memories somewhere. 

A second possibility, though, is that when we are reminded of the initial situation, our initial memory is actually opened up again in ways that allow it to be altered. That is, over time the initial memory may be gone completely and replaced with a revised version.

For a long time, the first of these possibilities was the one that was generally assumed by the field. More recently, though, studies suggest that our initial memories themselves may be changed in the future through a process called reconsolidation. In reconsolidation, a memory is made active again, and while it is active, it is subject to change.

One example of reconsolidation in people comes from a 2007 study by Almut Hupbach, Rebecca Gomez, Oliver Hardt, and Lynn Nadel published in Learning & Memory. They had participants study two lists of words over a three-day period. 

On the first day, participants learned a list of 20 words that named common objects.  They practiced the items until they could recall at least 17 of the 20 items. On Day 2, some participants were reminded that they had learned a list on the previous day. Others were not given a reminder. These two groups then learned a second list of words naming a different set of common objects. (A control group did not learn the second list.) On the third day, participants returned and were asked to remember as many of the words from the first list as possible.

The control group recalled about half of the words on the list. The group that was not reminded of the list that they learned on the first day recalled 45% of the words, and occasionally also recalled one of the words from the second list (about 5%). The group that was reminded of what they did on the previous day recalled only about 36% of the words from the first list. Interestingly, they also recalled about a quarter of the words from the second list they learned.

This finding suggests that just reminding people of the experience of learning the first list led them to combine their memory of the first list with that of the second. Two control conditions refined this finding a bit: In one study, participants recalled the first list immediately after learning the second list. In this study, participants did not recall any of the items from the second list when remembering those from the first. This finding suggests that it takes time for the memory of the second list to be combined with the memory of the first list.

Another control condition looked at memory for the second list. This study found that when people recalled the second list, they rarely added words from the first list to it, even when they had been reminded that they had learned the first list in the previous session. This study suggests that it is only the initial memory that is affected by a later experience.

Putting all of this research together suggests that it is possible to rewrite aspects of our old memories with new information acquired after the initial memory was created. These findings are particularly frightening when it comes to things like eyewitness memory, because it suggests that even if people were able to recall things correctly at some point in the past, that “truth” may no longer exist anywhere in memory.

This is just one reason why the legal system needs to treat eyewitness testimony carefully. After all, if old memories have been altered by new information, then the witness will still believe deeply in their testimony, because it reflects their actual memory. Unfortunately, that actual memory is not an accurate reflection of the past it represents.