The Misinformation Effect: How a Single Word Can Rewrite What You Remember

About this psychology article

Memory is often experienced as something closer to a video recording than it actually is. Decades of research show that a memory can be measurably altered after the fact, sometimes by nothing more...

A witness watches a short film of a car accident, then answers a series of questions about what she saw. One version of the question asks how fast the cars were going when they "hit" each other. Another version, shown to a different witness who watched the identical film, asks how fast the cars were going when they "smashed" into each other. The witnesses who heard "smashed" reliably estimate a higher speed than those who heard "hit," despite watching exactly the same footage.

That alone might be dismissed as a minor framing effect on a speed estimate, a fairly trivial detail to distort. But when the same witnesses were asked, a week later, whether they had seen broken glass at the accident scene, the "smashed" group was significantly more likely to falsely report seeing broken glass that was never actually in the film at all. A single word, embedded in a single question, had not just shaded an estimate. It had apparently altered what people believed they remembered seeing.

This is the core finding behind what's now called the misinformation effect, established through a series of experiments conducted by psychologist Elizabeth Loftus beginning in the 1970s, and it remains one of the most consequential findings in the psychology of memory, with direct, ongoing implications for how the legal system handles eyewitness testimony.

The Original Experiment

In the study described above, published by Loftus and John Palmer in 1974, participants watched films of staged traffic accidents, then were asked to estimate vehicle speed using one of several different verbs describing the collision: "contacted," "hit," "bumped," "collided," or "smashed." The specific verb used produced a clear, graded effect on estimated speed, with "smashed" producing the highest average estimates and "contacted" producing the lowest, despite every participant having watched an identical film.

The broken-glass follow-up, conducted as a second experiment, extended the finding considerably further. Since no broken glass actually appeared in any version of the film, any participant reporting having seen broken glass was demonstrably reporting a false memory, one that hadn't existed in their initial experience of the event at all. The "smashed" group falsely reported broken glass at roughly twice the rate of the "hit" group, providing direct evidence that a single word choice in a question, asked well after the actual event, could measurably distort not just an estimate but the actual content of a subsequent memory report. The full study, including its methodology and subsequent replications, is covered in The Misinformation Effect Explained .

The Broader Paradigm This Launched

Loftus's early word-choice experiments opened an entire research program studying what she termed the misinformation effect more generally: the finding that a person's memory of an event can be altered by exposure to misleading information encountered after the event occurred, whether through a leading question, a conversation with another witness, or media coverage discussing the event inaccurately. Subsequent research, using increasingly elaborate methodologies, extended this finding well beyond minor perceptual details.

In a particularly striking line of research, Loftus and colleagues demonstrated that entirely fabricated childhood events, including a documented study in which participants were led to believe they had been lost in a shopping mall as a child, an event confirmed by family members not to have occurred, could be implanted as apparently genuine memories in a meaningful proportion of participants, complete with added sensory and emotional detail the participants generated themselves once the false memory took hold. This research became known as the "lost in the mall" technique, and its implications extended the misinformation effect from distorting real memories to demonstrating that entirely false memories could be constructed from suggestion alone.

Why Memory Doesn't Work the Way Most People Assume

The misinformation effect research directly challenges what cognitive psychologists sometimes call the "video recorder" model of memory: the intuitive but inaccurate assumption that memory functions like a fixed recording, capturing an event accurately at the time it happens and simply replaying that fixed recording whenever it's later retrieved. The actual research suggests memory functions more like a reconstructive process, rebuilt somewhat freshly each time it's recalled, drawing on the original experience alongside whatever information, accurate or not, has been encountered in the interim.

This reconstructive quality means that each act of remembering is also, in a sense, an opportunity for the memory to be subtly altered, particularly when new, related information is introduced during the recall process itself. This connects to a related but distinct process covered in Cognitive Dissonance Explained , where the mind's tendency to maintain a coherent, consistent self-narrative can similarly reshape how past events and decisions get remembered and interpreted well after the fact.

The Specific Conditions That Make Memory More Vulnerable

Follow-up research identified several specific factors that increase susceptibility to the misinformation effect, which matter considerably for understanding when eyewitness testimony is most and least reliable. Time delay between the original event and the introduction of misleading information increases vulnerability, since the original memory trace weakens with time, creating more room for subsequent information to be incorporated. The perceived credibility and authority of the source introducing misleading information also matters considerably, with misinformation from a perceived expert or authority figure, including an interviewing police officer or attorney, showing stronger distorting effects than the same misinformation from a source perceived as less credible.

Additionally, research has found that misinformation is more likely to be incorporated when it's consistent with a person's existing expectations or schema for the type of event being recalled, since misleading details that fit a plausible, expected version of events face less resistance during the reconstructive recall process than details that would represent a more jarring departure from what seems reasonable.

Why This Matters for the Legal System Specifically

Loftus's research has had direct, substantial influence on how the legal system understands and handles eyewitness testimony, an area where the implications carry unusually high stakes. Eyewitness misidentification has been documented as a leading contributing factor in wrongful convictions later overturned through DNA evidence, and Loftus's research provided a clear, empirically grounded mechanism explaining how a witness's testimony, given in complete sincerity and confidence, can nonetheless be substantially inaccurate due to memory contamination that occurred somewhere between the original event and the eventual testimony.

This research has directly informed reforms in police interview and lineup procedures, including recommendations to avoid leading questions during witness interviews, to conduct lineups using officers unaware of who the actual suspect is, and to document witness confidence at the time of an initial identification, since confidence itself has been shown to be an unreliable indicator of accuracy, particularly once a witness has been exposed to any post-event information that could have contaminated their original memory.

Confidence Is Not Evidence of Accuracy

One of the more counterintuitive and important findings from this broader line of research is that a person's subjective confidence in a memory shows only a modest and inconsistent correlation with that memory's actual accuracy, particularly once misinformation has been introduced. Participants in Loftus's studies who had incorporated false details into their memory reports frequently expressed just as much confidence in the false details as in accurate ones, since from the participant's own internal experience, a successfully incorporated false memory feels indistinguishable from a genuine one.

This has significant implications well beyond the courtroom, since it suggests that the felt certainty accompanying any given memory, in any context, isn't reliable evidence of that memory's actual accuracy, a conclusion that runs against most people's strong intuitive sense that vivid, confidently held memories must be accurate ones.

What Actually Helps

Treating memory as reconstructive rather than fixed, particularly for high-stakes recollections. Recognizing that memory can be altered by information encountered after an event, especially during the period between the event and formal recall, supports more cautious handling of eyewitness testimony and personal recollection alike.

Avoiding leading or suggestive language when asking someone to recall an event. Given the demonstrated effect of even minor word choice on subsequent memory reports, open-ended, non-leading questions produce more reliable recall than questions that embed assumptions about what occurred.

Documenting recollections as soon as possible after an event. Since time delay increases vulnerability to misinformation, memories recorded close to the original event, before extensive discussion or media exposure, tend to be more reliable than those recorded after significant delay.

Treating high personal confidence in a memory with appropriate caution , particularly for memories that have been discussed, revisited, or exposed to related information multiple times since the original event, given the well-documented disconnect between subjective confidence and actual accuracy.

See How You Process and Recall Information Under Pressure

Understanding how memory can be reconstructed and distorted is a useful lens for examining your own patterns of recall, certainty, and self-narrative construction.

→ Take the Self Awareness Test

Free. No account required.

Want a broader look at your cognitive patterns, including how you process information under pressure?

→ Explore the full PsychOS Assessment Lab

Frequently Asked Questions

What is the misinformation effect?

The misinformation effect describes how a person's memory of an event can be altered by exposure to misleading information encountered after the event occurred, whether through leading questions, conversations with other witnesses, or media coverage, first demonstrated systematically by psychologist Elizabeth Loftus in the 1970s.

Can completely false memories actually be implanted in someone?

Research suggests yes, to a meaningful degree. Loftus's "lost in the mall" studies demonstrated that entirely fabricated childhood events could be implanted as apparently genuine memories in a portion of participants, complete with added sensory and emotional detail generated by the participants themselves.

Why does this matter for eyewitness testimony in court?

Eyewitness misidentification is a documented leading factor in wrongful convictions later overturned by DNA evidence. Loftus's research provided a clear mechanism explaining how sincere, confident testimony can still be substantially inaccurate due to memory contamination occurring after the original event.

Does feeling confident about a memory mean it's accurate?

Not reliably. Research has found only a modest and inconsistent correlation between subjective confidence and actual memory accuracy, particularly after misinformation has been introduced, since a false memory can feel just as vivid and certain as an accurate one.

How can memory distortion be reduced in practice?

Avoiding leading or suggestive language when asking about an event, documenting recollections as soon as possible after they occur, and maintaining appropriate skepticism toward highly confident memories that have been revisited or discussed multiple times all help reduce vulnerability to the misinformation effect.