Crash Into Me(mory): Memory Is an Active, Reconstructive Process

Pull out a piece of paper and jot down how much money you made this year. Now, write how much money you made two years ago. All set? Now here's the fun part. Go find your W-2 tax forms. How accurate were you? I'm going to guess that you were fairly accurate in estimating both numbers. However, if you were off, even by a little bit, then it was probably in a reliable direction. If you made more money this year than you did two years ago, then I bet your estimate of your income two years ago was a little bit on the high side. If you make less than you did two years ago, then I'm willing to bet that your estimate was a little low. 

Not having access to your IRS files, why would I bet that you were off? And how could I know what direction your error was in? As it turns out, memory errors are somewhat predictable [1]. You might be thinking to yourself: Is memory fallible? (Short Answer: Yes, it is.) Are errors in memory reliable and/or predictable? (Short Answer: Yes, they are.) Our goal in this post is to figure out how and why.

Do people change?

Most people have a theory about how individuals change as they mature and grow older. A person's theory of change can be one of stability. For example, some might assume that their political beliefs are relatively stable over time and do not change from one election year to the next. Theory of change can also be dynamic in the sense that people believe they are essentially different people from when they were young. As a kid, for instance, you might have believed that the sun revolved around the earth, but then you learned in science class that the earth actually revolves around the sun.

What does one's theory of change have to do with memory? Above I made the claim that memory is predictably fallible [2]. In fact, memory is an active, reconstructive process where the individual uses her current situation to help unpack and understand past events (i.e., episodic memories). The implication is that memory is not a stable entity like a video recording. Instead, it is full of inferences based on two things: what's going on in the present moment and one's theory of change. 

Remembering how much money you made two years ago is a perfect example. When trying to recall the exact figure, you first anchored your estimate on your current situation (e.g., How much do I currently make?). Then you invoked your theory of how your income changed (e.g., Did I get a raise or take a pay-cut?). Based on these two data points, you can then "recall" (or reconstruct) how much money you made. In other words, the memory of the exact figure is highly influenced by your current situation and your theory of change with regard to your income.

"I'd like to voir dire this witness..." -V. Gambini

One of the best demonstrations of the inferential nature of memory can be found in the literature on eyewitness testimony. When a crime is committed, and there is no photographic evidence, we must rely on eyewitnesses to testify as to what they saw. Unfortunately, memory does not act like a video recording where the memory replays exactly the same every time. Instead, how we remember something can change based on our current situation.

Given the flexibility of the English language, lawyers can influence how an eyewitness remembers something by framing their questions in different ways:

   Framing 1: "About how fast were the cars going when they hit each other?"

   Framing 2: "About how fast were the cars going when they smashed into each other?"

If memory acted like a video recording, then it shouldn't matter how the lawyer asks her question. The eyewitness should provide similar estimates of the car's speed. However, if memory is an active, reconstructive process, then the estimates should change depending on the way the question is framed. 

To find out if the question framing had an effect on memory, a pair of scientists invited a group of participants to watch a video of a car crash [3]. After a delay, participants were then asked to estimate the speed of the car. Based on the two framings above, what do you suppose they found? Participants who heard the question framed using the word "hit" estimated that the car was traveling 34.0 mph. However, participants who heard the question framed in terms of "smashed into" estimated that the car was traveling 40.5 mph. Even though both groups of participants saw the exact same video, their memory of the event was highly influenced by the wording of the question.

The STEM Connection

Why is this important for education? First of all, I believe that it is important for students to have an accurate understanding about the way their memory works. If they believe the misconception that memory is like a video recording, then that belief will guide their behavior. For instance, they will be more likely to believe their own memories of past events and may also trust the memories other people share with them to be hard facts. But if they understood how memory can be biased by the current situation and an individual's theory of change, then they might be more skeptical of the recalled information. They might also study differently if they understand that memories are error prone. 

Second, it seems reasonable to assume that students might develop better metacognitive skills if they have an accurate understanding about the way memory works. In other words, if they understand that memory is fallible, then they might try various technique to reduce the possibility of introducing errors into their episodic memories. They might not rely on memory at all! Instead, they may elect to store episodic memories externally (e.g., in pictures or in a journal). 

To summarize, remember that memory is fallible, and that it is not like a video or audio recording. The present is typically used as an anchor and inferences about the past need to be made. Also, the next time you're called for jury duty, pay attention to the way lawyers ask their questions. It could make serving jury duty a smashing good time.


Share and Enjoy!

Dr. Bob

For More Information

[1] Ross, M. (1989). Relation of implicit theories to the construction of personal histories. Psychological Review, 96(2), 341.

[2] Methodologically speaking, the claim that memory is predictably fallible is difficult to verify scientifically. To test this claim, you would need to follow someone around and record their beliefs, and then wait for an extended period of time and ask them to recall what you had recorded. Then you would have an accurate measure of the truth and any deviations in memory when recalling the event. That typically isn't practical, so researchers need to be creative in collecting autobiographical data and testing the accuracy of those individual's memories. One way to get around this methodological hurdle is to ask people to keep a journal. Assuming the person faithfully keeps track of the events in his or her life, then memories can be compared against the journal entries.

[3] Here is a replication of the video used in the experiment, and here is the paper where they report the results: Loftus, E. F., & Palmer, J. C. (1974). Reconstruction of automobile destruction: An example of the interaction between language and memory. Journal of verbal learning and verbal behavior, 13(5), 585-589.