Do you have a friend who tells the same story at every party? Have you noticed how on each iteration the sky gets darker, the water gets deeper, and the guppy slowly morphs into a whale? Does this mean your friend is a big fat liar? Well maybe a little. But, this is also a natural characteristic of our memory. It turns out our memories are not static representations of past events like a photo or a video recording, but they dynamically change every time we think about them.
It may seem troublesome that we don’t keep accurate representations of the past stored in our memories -- but I would argue that this is actually an adaptive process that allows us to cope with our constantly-changing situations. For instance, if our memories stayed the same and never changed, you would never learn that your partner is now 20 years older than when you met him, because the younger version of him would prevail in your memory — it would never get updated with new information (gray hair, wrinkles around the eyes, etc.).
Remember the Telephone Game? The first person whispers “My new blue shoes”. After a few whispers down the line, the final person hears “A new bus on the moon”. Each person hears something a little different, because they only receive a distorted version of the original due to low-quality signal. In the telephone game, some words drop out entirely from the message, other words get changed to new words, and completely new words somehow get inserted out of thin air. Our memories behave similarly. In the telephone game, one message is passed from one person to another. But with memory, the message is passed from our past self to our current self every time we remember it. And this happens over and over again each time we remember an experience.
Contrast this idea with how people usually think about memory: as some unchanging photograph. In this view, when you remember an event, you take out your “original” or “master copy” of the photograph, and look at it. You might notice different things about the photo, but it’s still the same, unchanging picture. The “telephone game” theory of memory on the other hand has no master copy, it just has a picture that is continuously altered every time you check it out. It's like the photo gets photoshopped every time you look at it. Maybe you remove those pesky wrinkles or cut out that annoying person in the background photobombing you and your friends. Each time the photo is altered, it get's a little further from reality.
This feels a little weird. Why wouldn’t we notice our memories changing? It’s because we only experienced the event once, but we recall it many times. Unfortunately, unless we had a video recording of the entire event (and a view from our first person perspective), we can’t compare our current version to the original. We only have the last time we remembered it fresh in our memories. Because of that, we can be highly confident that we are accurately remembering an event, when in fact, we are recalling the last time we remembered it.
Our memories can change in a few different ways. First, the most common complaint about memory is that we forget details (and sometimes entire events). What was his name? What color dress was she wearing? Second, we don’t just forget, but we fill in pieces of the story with information that was not from the original experience. The filler information is often something we've encountered, just not during the original event. This is like two different experiences crossing over into one memory. It might seem relatively inconsequential to forget some details or for some details to morph into others, but it’s disconcerting how this type of information is relied heavily on in a courtroom when there is an eye witness.
Most memory studies rely on showing someone a stimulus (like a word or photograph), some waiting, and then asking the person if they remember the stimulus. This yields a simple “yes” or “no” response -- there is no gray area. It’s difficult to measure how much a memory changed in this type of test format. To get around this problem, I designed a spatial memory task that required subjects to use the mouse to place an object back in its location. This task yielded a graded measure of memory: distance. I was able to measure the distance each object was placed from the original location and track how placement changed over the course of multiple tests.
I designed the experiment to examine how a memory shifts each time we remember it. In this study, participants learned object-location associations, meaning each object was presented in a specific location on the screen. I tested the subjects every day for three days. Subjects were asked to place each object back in its original location, using a computer mouse to drag the object to its spot on the screen. As in real life after we experience an event, subjects never received feedback about whether or not they had placed the object in the correct spot. With this test, I was able to measure the two features of memory distortion: 1) How much did participants forget the original location of the course of the three days? and 2) Did the locations they chose depend more on the original location or on the location that they had last selected?
First I looked at forgetting by seeing how far away they placed the object from the original location over the 3-day period. For this analysis, I measured the distance each object was placed from its original location on each test at Day 1, Day 2, and Day 3. Interestingly, there was a big drop in performance from Day 1 to Day 2, but forgetting leveled off, as memory declined substantially less from Day 2 to Day 3. These results suggest that after an event, our memory drops off immediately, and we forget a lot of the details and context information. But after that initial drop-off period, our memories maintain stability. These results are in line with other studies of memory consolidation that show that older memories are more stable than newer memories. It also resembles the classic Ebbinghause forgetting curve, which shows a steep drop-off in accuracy immediately after learning and then levels off as time goes on.
Next I looked at how each test influenced the content of memory on each subsequent test. For this analysis, I looked at final memory outcome on the Day 3 test. I measured the distance each object was placed from 3 different locations: the originally studied location, the location subjects selected on the Day 1 test, and the location subjects selected on the Day 2 test. In this way, I could see how each of these experiences influenced where subjects placed the object on the final Day 3 test. Interestingly, the results showed increasing influence of the most recent test on final memory performance. Subjects placed objects closest to the Day 2 test location, followed by the Day 1 test location, and farthest from the originally studied location (Bridge, 2012).
These results show the powerful influence each experience of remembering has on our final memory. Each time we remember an event, information from that experience gets incorporated into the original memory, causing it to change slightly each time. This type of study makes you wonder what a memory actually is. How much is it really reliving the original experience? It seems that our memories are more colored by each experience we have of reminiscing about those old times. Our memories change with us, and reflect more about who we are rather than who we were.
Based on this study, I hypothesized that every time we remember an event, it is changed by incorporating the retrieved information into the original memory. But, there is an alternative hypothesis. It could be that we are just more likely to remember things that happened recently rather than things that happened far in the past. It is possible that subjects remembered the Day 2 test location the best on the Day 3 test, because they encountered that location most recently. To get to the bottom of this, I designed another study to determine if recency was really the underlying cause of this updating effect.
In this new study, I had subjects encounter some incorrect information about an event they previously experienced. However, I manipulated how subjects encountered this incorrect information. In one condition, I had subjects try to recall the original event, similar to the first study. In a second condition, I showed subjects incorrect information about the original event (but i didn't tell them it was incorrect).
With these two conditions, I could test the hypothesis that remembering events is really a special time to update our memory. If the updating effect I saw in the first study was really just about recency, it shouldn’t matter if I show subjects the wrong location or if they mistakenly recall the wrong location -- they should remember the most recently encountered location on the final test. On the other hand, if updating only happens when we actively remember the wrong location, then subjects should only choose the updated location in the Active condition, but not the Passive condition.
The study design was similar to the first study, but I introduced a new condition. Just like the first study, subjects studied objects in unique locations on the screen. Then, in an “Active” condition, I asked subjects to place each object back in its original location. In a new "Passive" condition, I chose a new location and instructed subjects to place the object in that spot (the yellow box indicated where subjects should place the apple). In both conditions, subjects placed the object in the wrong location. The main difference is that in the Active condition, subjects had to try to remember the object's location (and they always got it a little wrong), whereas in the Passive condition, I chose the wrong location for subjects to place each object in. After a delay, subjects completed a final test. On this test, subjects saw each object in 3 locations: the original location, the updated location, and a brand new location. Subjects attempted to select the object's original location.
It turns out remembering events is really special. Subjects were not easily tricked into thinking the wrong location I selected was the object’s original location. In the Passive location, subjects disproportionately chose the original location over the other options. However, updating was prevalent in the Active condition. Subjects usually chose the updated (incorrectly remembered) location on the final test instead of the other choices. These results show that remembering events provides an opportunity to update them. The good news is that we are not so easily fooled by incorrect information that does not match our memories (Bridge, 2014).
During this memory task, we also collected brain scans of subjects using fMRI to identify the neural correlates associated with memory updating in the Active condition and memory stability in the Passive condition. fMRI measures oxygenated blood flow (see Consciousness Continuum post), which is a proxy for neural activity. The main advantage of fMRI is that it can tell us which parts of the brain contribute to specific memory processes.
The primary fMRI results revealed activity in the hippocampus, which is the brain region responsible for forming and remembering memories of specific events. Our results showed that activity in the hippocampus was associated with memory stability in the Passive condition and memory updating in the Active condition.
This fMRI activity suggests that the same neural mechanism involved in remembering real events is also involved in remembering updated events. We infer that the hippocampus builds memories based on whatever is most adaptive: updating happens during remembering, so that new, relevant information can be incorporated into the original memory. On the other hand, memory stability happens when we are presented with inaccurate information, so that the original memory is protected against erroneous information that we may encounter. Importantly, these results suggest that both memory change and stability are fundamental memory processes that occur fluidly and dynamically over the life of a memory.
These studies involved explicit responses from participants. On each test, I explicitly asked participants to place the objects back in their original locations. On the final test, they tended to erroneously remember the previously remembered location rather than the original location. I would argue that the process of memory updating was unconscious. Subjects were not aware that their memory changed each time I tested them. These studies go to show that even when we do some explicit action (intentionally remembering an event), new information can seep into that memory due to unconscious processes happening in the brain.