True memories fade and false ones appear.
Each time we recall something, the memory is imperfectly re-stitched by our brains. Our memories retain familiarity but, like our childhood blankets, can be recognizable yet filled with holes and worn down with time.
To date, research has shown that it is fairly easy to take advantage of our fallible memory. Elizabeth Loftus, cognitive psychologist and expert on human memory, has found that simply changing one word in a question can contort what we recall. In one experiment, Loftus had participants watch a film of a car crash, and then asked about what they saw. They were either asked “How fast were the cars going when they hit each other,” or “How fast were the cars going when they smashed into each other.” One week later the participants returned for some memory questions. Loftus asked whether or not there was broken glass at the scene of the accident. Those participants that heard the word “smashed” were more than twice as likely to recall seeing broken glass than those who heard the word “hit.” Keep in mind, there was in fact no broken glass at the scene.
This kind of insight—that our memories are terrible camcorders of reality—had serious pop culture ramifications. “Repression” and “repressed memories” have entered our culture’s lexicon, without evidential support. Even with numerous accusations of sexual abuse and other childhood horrors filed in court with the explosion of “recovered memory therapy,” the same research pioneered by experts like Loftus has suggested that most if not all of these “repressed” memories are merely false ones. At CSICon, a skeptic’s conference earlier this year in Nashville, Tennessee, Loftus herself noted that the same techniques used to implant false memories in psychological experiments are precisely the techniques used by repression therapists to recover supposedly buried traumas.
Nearly four decades later, Loftus and colleagues aim to further memory science once again. Introducing a false memory in experiment can be done quickly and with some degree of reliability, but how long does the lie last? Surely bolstered by a digital age reverberating with misinformation, the results point to a disturbingly long half-life of lies.
Earlier this year, Zhu et al. tested the veracity of a quickly incepted false memory. After selecting 342 participants, the research team set about twisting their memory with two events, shown on slides, crafted to encourage endorsement of a falsehood. The researchers showed participants 50 slides of each event in quick succession, one depicting a man breaking into a car and stealing things from it, and the other depicting a girl’s wallet being stolen by a seemingly nice man.
To create a false memory, the researchers followed this slideshow with a narration of the events that took place, mimicking “eyewitness” accounts. 50 sentences were presented to participants that supposedly accurately described each slide seen in the events. However, 12 “key” slides were manipulated in the narration sentences. For example, if participants saw the thief put the woman’s wallet in his jacket pocket, the narration would describe the thief putting the wallet in his pants pocket. Other participants would see different combinations of these memory manipulations.
The research team then tested for the implantation of false memories by giving participants both a recognition test (what did you see in the pictures?) and a source-monitoring test (did you choose that answer because you read it, saw it, both, or guessed?).
Importantly, the researchers made sure that false memories could be separated from simply wrong ones. For example, the slides would show a man stealing a woman’s wallet and then hiding behind a tree, while the narration would describe him hiding behind a door after the petty theft. When asked, “where was the man hiding?” in the recognition test, participants could answer either “behind a tree” (true memory), “behind a door” (false memory), or “behind a car” (wrong/“foil” memory). In this way, Zhu et al. could determine if their false memories incepted by the narrations stuck in the participants’ minds.
A year and a half later the same event slides were presented to the original participants. But this time during the slideshow, the experimenters paused it and asked participants what would happen next, right before the “key” slides from the original test. What was remembered from the testing 1.5 years prior would the participants’ only way to advance the slideshow.
Disturbingly, as the researchers note, “the false memory briefly introduced in an experimental setting seemed to have similar strength…as true memory” (pg. 306).
In the first test, participants endorsed 61% of the true items and 31% of the misinformation items as what really happened in the 12 manipulated slides. A year and a half later, only 45% of the true items were endorsed, showing a decent memory. But this time, 39% of the misinformation items were taken as true, a statistically significant increase.
We seem to think that misinformation is somehow “weaker” than the truth, that it does not last as long or pierce as deep. The research disagrees. Misinformation can be just as enduring, and even increase in strength over time. This has real world consequences. Consider how critically we must view eyewitness testimony (even more so than usual) when there is conflicting video and verbal accounts. Maybe a policeman has a witness watch a burglary caught on tape, pausing the video to ask what happens next. Distinguishing between a true and a false memory based on strength of recall then seems a futile exercise.
True memories fade, and lies darken.
Sharing is now one of the easiest things to do, given a decent bandwidth. And rumors spread like wildfire. Numerous celebrities have been declared dead by social media and reported in the mainstream (Jeff Goldblum, Bill Nye, Morgan Freeman, etc.). A group of bloggers nearly “screwed” the entire Apple community with a simple lie, and documented just how expansive it became.
Most recently, the landfall of hurricane Sandy generated a steady stream of lies in the form of doctored pictures. Unnatural clouds loomed over New York, the Statue of Liberty was engulfed by a monstrous “Day After Tomorrow” wave, and sharks patrolled the streets of New Jersey. Though sources like The Atlantic provided real-time debunking of many of these images, I suspect that the fake photos made much more of a dent than their corrections.
So in a deluge of Sandy tweets and links and articles we witnessed a slightly flipped version of the memory study discussed above. Fake images were presented, corrected shortly after by narrations (and further pictures) from news websites and knowledgeable tweeters. The discrepancy between what was seen (“A shark on a New Jersey lawn!”) and what was read (“That shark photo was a fake!”) made for a perfect experiment in implanting false memories.
The data on this disconnect would need to be collected and crunched, but based on memory science we know what to expect. Years from now, when we think back on hurricane Sandy and the destruction it caused, we are eventually bound to hear:
"Remember when sharks were swimming the streets of New Jersey?"
Figure from Zhu, et al. 2012.
1. Brandon S, Boakes J, Glaser D & Green R (1998). "Recovered memories of childhood sexual abuse: implications for clinical practice". British Journal of Psychiatry 172: 296–307.
2. 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, 585-589.
3. Zhu, B., Chen, C., Loftus, E. F., He, Q., Chen, C., Lei, X., et al. (2012). Brief Exposure to Misinformation Can Lead to Long-Term False Memories. Applied Cognitive Psychology, 301-307. Published online 15 December 2011 in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/acp.1825