Can we trust our memory?

July 15th, 2025

Written by: Victoria Subritzky Katz

It’s easy to think of memory like a video camera, recording the events of our life and storing them safely in the brain so that later we can ‘hit play’ on the stored recording to see what happened. But science tells a very different story. Memory can be surprisingly fragile— and its imperfections can carry serious consequences, especially in the courtroom. In this piece, we will explore how memory works, how vulnerable it is to outside influence, and how it can still play a role in legal proceedings if handled correctly.

How Memory Works

While our brain doesn’t record the events of our lives as perfectly as a camera, we can understand the steps to make a new memory in similar terms. First, an experience is encoded as a new memory, like recording new footage, before it is then consolidated in the brain, similar to storing a video recording on a disk drive. Later, when you remember that experience it is retrieved from its storage location in the brain, like playing back the old footage.

However, unlike a video camera, inaccuracies can creep in at each of these stages. When encoding a memory, our brain doesn’t capture every single detail. Instead, it focuses on elements meaningful to us, often leaving gaps or inaccuracies caused by omissions of information. For example, Figure 1 shows how the banana is not encoded in memory, perhaps due to a preference for apples or because the bug makes the apple more interesting, leading to an inaccuracy in the memory caused by this omission.

When consolidating a memory in the brain it mixes with the information already stored there, including our past experiences and biases, which can lead to memory inaccuracies caused by contamination. For instance, in the fruit memory example, the color of the apple changes to red when consolidated due to contamination from past experiences or a bias that apples are often red.

Lastly, subsequent retrievals of the same memory can cause further inaccuracies. Each act of remembering makes the memory vulnerable to influence from new experiences and the context in which it is recalled, leading to modifications which then become part of the memory.The next time we recall the event, we are not remembering the original event, but the modified version. In the fruit example, the single ant on the apple gets modified to two after multiple remembrances^4,5.

A memory’s sensitivity to corruption from these inaccuracies of omission, contamination, and modification has led scientists to conclude that memory is far from set in stone. Instead, it is fluid and always evolving, making it vulnerable to subtle shifts and distortions we may not even notice. But just how vulnerable is our memory to tampering from outside influence?

False Memories: Remembering What Never Happened

While we all know what it feels like to forget details or mix up timelines, it’s more unsettling to realize that we can also form entirely false memories. Researchers have demonstrated they can subtly influence participants to believe an event occurred in their childhood that never happened, through a process referred to as false memory implantation.

In memory implantation experiments, adult participants are told the study is about childhood memories and are presented with several events that occurred during their childhood, as reported by their parents or siblings. However, one of these events is a fabrication, entirely made-up by the researcher. The fabricated event is designed to be unique and memorable, such as going on a hot air balloon ride or spilling the punch bowl on the bride’s parents at a wedding. Over the course of the week, the participant is interviewed 2-3 times and asked to recount each memory (including the fabricated memory they are unaware is false).

Having to repeatedly remember an event opens the door for the type of memory modifications discussed in the first section, where the original details (in this case the absence of) are replaced by the version most recently recollected. Combined with the social pressure of being told a family member remembers the event, researchers can successfully persuade many participants that the fake event was a real part of their childhood^1,7.

This may sound like science fiction— can a few conversations really make someone believe they once floated above the ground in a basket dangling below a giant balloon? Yet the results of these studies show that implanting false memories in our mind has leaped from the pages of fiction books into reality.

One study that compared the effects of memory implantation in 423 participants across eight smaller studies found about one-third of people fully accepted the false memory and more than half showed some evidence of believing the event occurred⁷. To be classified as a “false memory,” participants had to both accept the event as real (for example, saying unprompted, “I remember…”) and add new details or imagery not originally provided. Importantly, this was assessed during a separate final interview, after the earlier sessions in which researchers actively tried to implant the memory.

While not everyone was convinced, these findings illustrate that memory is not always as reliable as we might hope. In our daily lives our memory is unlikely to be purposefully tampered with in this way, but it does show how vulnerable a memory can be to outside influence and manipulation. 

Memory on the Witness Stand

Alarmingly, one place where memory can be bombarded by outside perspectives and face strong external pressure is in the courtroom. The vulnerability of memory becomes especially critical when eyewitness testimony can make or break cases.

During criminal investigations, witnesses face a host of potentially corrupting influences: suggestive questions from investigators, new information presented about the case, and pressure (external or internal) to recall events in a way that aligns with a specific narrative. As we’ve discussed, memory is not set in stone, and in addition to the inaccuracies that can slip in during the encoding and consolidation stages, each retrieval of a memory makes it vulnerable to modification. This makes the need to recall memories multiple times— in interviews, lineups, or in court— the most damaging to a memory’s reliability as it opens the door to these corrupting influences.

One way this is apparent is in the ‘strengthening’ of an incomplete or hazy memory after repeated recollections. Studies have shown how a witness who starts out feeling unsure about identifying a suspect (rating their confidence as a shaky 3 out of 5), can grow more assured of their belief after multiple lineups and repeated questionings, despite evaluations of their accuracy in research settings remaining the same⁹. By the time they testify in court, they may firmly believe in a memory that has been subtly reshaped and reinforced over time.

The consequences of the vulnerability of memory when used as legal evidence can be profound. Eyewitness misidentification played a role in roughly 70% of the 350 wrongful convictions that have been overturned by DNA evidence thanks to The Innocence Project⁹.

Should We Trust Our Memory?

Given these vulnerabilities, some experts have called into question the overall reliability of memory and recommended the discounting or exclusion of memory-based evidence in legal proceedings^1,4,9. However, while our memory is not as picture-perfect as we might hope, it can still be largely trusted as long as we recognize and account for its limitations.

Like DNA evidence, which can be contaminated if mishandled, the integrity of memory can be preserved through proper safeguards. In court, this means using open-ended questions during interrogations, relying mainly on a witness’s first recollection, and considering their initial confidence as a measure of memory strength^3,9. Relying on a witness’s first recollection helps preserve accuracy by avoiding the distortions introduced through repeated retrievals. The Innocence Project cases illustrate this clearly: for all the wrongful convictions based on eyewitness testimony (and for which data were available), the witness was very uncertain during their initial suspect identification despite their later strong confidence in trial. The witness’s memory only became inaccurate after repeated recalls, an error that could have been avoided by relying on the initial, less contaminated account.

It’s important to understand the limits of memory, whether we’re recalling everyday moments or deciding someone’s fate in court. Memory isn’t set in stone, but that doesn’t mean it should be dismissed. Instead, we need to respect its imperfections and handle it with care, especially when it guides judgments of guilt or innocence. Our memories may not be perfect, but they don’t need to be discarded — they just need to be used wisely.

References

1. Arce R, Selaya A, Sanmarco J, Fariña F. Implanting rich autobiographical false memories: Meta-analysis for forensic practice and judicial judgment making. Int J Clin Health Psychol. 2023 Oct-Dec;23(4):100386. doi: 10.1016/j.ijchp.2023.100386. Epub 2023 Apr 15. PMID: 37113566; PMCID: PMC10126919.
2. Brewin, C. R., Andrews, B., & Mickes, L. (2020). Regaining Consensus on the Reliability of Memory. Current Directions in Psychological Science, 29(2), 121-125.
3. Conway, A.R.A., Skitka, L.J., Hemmerich, J.A. and Kershaw, T.C. (2009), Flashbulb memory for 11 September 2001. Appl. Cognit. Psychol., 23: 605-623. 
4. Howe ML, Knott LM. The fallibility of memory in judicial processes: lessons from the past and their modern consequences. Memory. 2015;23(5):633-56. doi: 10.1080/09658211.2015.1010709. Epub 2015 Feb 23. PMID: 25706242; PMCID: PMC4409058.
5. Lentoor AG. Cognitive and neural mechanisms underlying false memories: misinformation, distortion or erroneous configuration? AIMS Neurosci. 2023 Sep 7;10(3):255-268. doi: 10.3934/Neuroscience.2023020. PMID: 37841346; PMCID: PMC10567586.
6. OpenAI. (2024). ChatGPT (July 2024 version) [Large language model].
7. Scoboria, A., Wade, K. A., Lindsay, D. S., Azad, T., Strange, D., Ost, J., & Hyman, I. E. (2016). A mega-analysis of memory reports from eight peer-reviewed false memory implantation studies. Memory, 25(2), 146–163.
8. Talarico, J. M., & Rubin, D. C. (2003). Confidence, Not Consistency, Characterizes Flashbulb Memories. Psychological Science, 14(5), 455-461.
9. Wixted, J. T., Mickes, L., & Fisher, R. P. (2018). Rethinking the Reliability of Eyewitness Memory. Perspectives on Psychological Science, 13(3), 324-335.

Cover photo from FreePik.