Recall
How much can we actually trust our memories?
What happens when something as personal as a human memory becomes part of a criminal investigation?
Hi, my name is Alia, and I’m still in my freshman year of school. I know that might make what I’m about to talk about sound like a big subject for someone my age. I’m not a psychologist, a lawyer, a police officer, or a neuroscientist. I’m a student who is curious about how the world works and how ideas from different fields could eventually be used to help people.
I’ve always found psychology interesting, especially when it comes to understanding why people behave the way they do and how our minds influence the decisions we make. But while exploring the subject, I came across something that made me look at memory in a completely different way.
Have you ever remembered something with complete confidence, only to find out later that one of the details was wrong? Or heard someone describe an event differently from how you remember it, even though you were both there?
It made me wonder how much we can actually trust our memories.
At first, this might seem like a simple question. We remember things every day without thinking twice about it. However, memory becomes a much more serious subject when it is used to make decisions about other people. In a criminal investigation, for example, one person’s recollection could influence the direction of a case. If that recollection is inaccurate, the consequences can be significant.
Someone’s freedom, reputation, family, education, and future could be affected by what another person believes they remember.
That was the starting point for RECALL.
From psychology to eyewitness memory
My research began with psychology, which gradually led me towards cognitive psychology. This area focuses on the mental processes involved in things such as attention, perception, learning, and memory.
The more I explored memory, the more I became interested in false memories. I found it fascinating that someone could genuinely believe they remembered an event, even when parts of that recollection had changed over time.
This led me to eyewitness memory and its role in criminal investigations.
Eyewitness testimony can be an important part of an investigation. A witness may have seen something happen and be able to describe what they remember. However, being certain about a memory does not necessarily mean that every detail is accurate.
The consequences of this become clearer when looking at real cases.
According to the Innocence Project’s historical dataset, 375 people in the United States had been exonerated through DNA evidence between 1989 and 2020. Of those cases, 69% involved eyewitness misidentification. Together, these individuals had spent 5,284 years in prison, with an average of approximately 14 years before being exonerated.
These statistics do not mean that eyewitness testimony is always unreliable or that every wrongful conviction happens because of a mistaken memory. They do show how serious the consequences of misidentification can be.
What interested me most was that a witness does not necessarily have to be lying for their recollection to be inaccurate. They may honestly believe what they are saying. The problem is that human memory is not always as reliable as we assume.
The science behind human memory
To understand why memories can change, I first needed to understand how the brain forms and retrieves them.
Memory is usually explained through three main processes: encoding, storage, and retrieval.
Encoding is when information is first taken in and processed by the brain. Storage refers to keeping that information over time, while retrieval is the process of bringing it back when we try to remember something.
Although these stages sound quite straightforward, memory is much more complicated than simply recording an experience and playing it back later.
Our brains do not store every moment like a video. Instead, remembering involves reconstructing information using details from the original experience, alongside our existing knowledge and anything we may have learned afterwards.
This is known as constructive memory.
Research by Daniel Schacter, Kenneth Norman, and Wilma Koutstaal explored how the brain constructs memories and why this process can sometimes lead to errors. Their work helps explain why remembering is not always an exact reproduction of the past.
Another concept worth understanding is memory reconsolidation. Under certain conditions, retrieving an existing memory can make it temporarily more open to modification before it is stabilised again. The exact conditions and processes involved in human memory are still being researched, so this should not be treated as something that happens every time we remember an event.
What matters for this project is that memories are not necessarily fixed. They can be influenced by different experiences and information, even after the original event has taken place.
When a question becomes a suggestion
One of the most interesting things I discovered was the misinformation effect. This happens when information introduced after an event influences how someone remembers that event.
A well-known study on this subject was conducted by psychologists Elizabeth Loftus and John Palmer in 1974.
In their experiment, participants watched videos of car accidents. Afterwards, they were asked questions about what they had seen. The researchers changed the wording of one particular question, asking some participants how fast the cars were going when they “smashed” into each other, while others received questions using less forceful wording.
The results showed that the words used in the question affected how fast participants estimated the cars had been travelling. When participants were questioned again later, those who had heard the word “smashed” were also more likely to report seeing broken glass, even though there had been no broken glass in the original video.
This study showed how information introduced after an event can become part of someone’s later recollection.
It also made me think about the way witnesses are questioned during investigations. The wording of a question might seem like a small detail, but it can influence the way someone interprets or remembers an experience.
Another interesting example is the verbal overshadowing effect.
In a 1990 study, psychologists Jonathan Schooler and Tonya Engstler-Schooler investigated whether describing a face in words could affect someone’s ability to recognise it later.
Participants watched a video containing a particular person’s face. Some were then asked to describe the face, while others were not. When they were later tested on their ability to recognise the person, those who had described the face performed worse than those who had not.
The researchers suggested that putting a visual memory into words could sometimes interfere with the way the original visual information was accessed.
This does not mean that describing a memory will always make it less accurate. It shows that even the process of explaining what we remember can sometimes affect how we retrieve it.
Both studies helped me understand that memory can be influenced by more than the original experience. Questions, conversations, and the way we describe an event can all play a role.
The problem of source monitoring
There is another part of memory that I found particularly relevant to eyewitness testimony: source monitoring.
Source monitoring is our ability to recognise where a particular piece of information came from.
For example, imagine someone remembers seeing a particular detail at the scene of an event. However, they may have actually learned about that detail later from another witness, a photograph, or something they read online.
They remember the information, but they may not remember where it originally came from.
This is an important distinction because remembering a detail and remembering its source are not necessarily the same thing.
Psychologists Marcia Johnson, Shahin Hashtroudi, and D. Stephen Lindsay explored this concept in a paper published in 1993. They explained how people use different features of a memory, including its context and perceptual details, to work out where the information originated.
However, these judgements can sometimes be incorrect. Details from different sources can become difficult to separate, particularly when a long period of time has passed.
In an eyewitness situation, this could mean that someone genuinely remembers a detail but mistakenly believes they witnessed it themselves.
This made me realise that investigating memory requires more than simply asking someone what they remember. It also means considering how they formed that recollection and where the information may have come from.
Why are some people more resistant?
Something else I wanted to explore was whether everyone is equally likely to be influenced by misleading information.
Research suggests that differences in cognitive abilities may affect how vulnerable someone is to memory distortion.
A particularly useful study for this part of my research was a systematic review published in 2024 by Maryanne Brassil, Cian O’Mahony, and Ciara M. Greene. It examined whether cognitive abilities could reduce a person’s susceptibility to the misinformation effect.
The researchers reviewed nine studies and identified 23 associations between cognitive abilities and susceptibility to misinformation. They considered three main areas: general intelligence and reasoning, perceptual abilities, and memory abilities.
Across the studies, stronger cognitive abilities were often associated with lower susceptibility to misinformation. However, this did not mean that people with stronger abilities were completely protected from memory distortion.
I wanted to understand why these abilities might make a difference, so I looked at how they relate to the brain.
Reasoning and decision-making involve several brain systems, including areas of the prefrontal cortex. These systems help us evaluate information, recognise inconsistencies, and control our attention.
Perceptual abilities are also important because they affect how accurately we process information in the first place. If someone does not properly notice an important detail during an event, they may struggle to remember it later.
Working memory is another relevant ability. It allows us to temporarily hold and work with information. This could help someone compare an original recollection with information they encounter afterwards.
The hippocampus also plays an important role in forming and retrieving episodic memories, which are memories of personal experiences. It works alongside other brain regions involved in attention, decision-making, and monitoring the source of information.
These processes do not happen independently. They work together as part of a wider network.
One way to understand this is to consider the stages involved in remembering an event:
Perception, encoding, maintaining information, comparing details, identifying the source, and retrieving the memory.
If something changes or goes wrong at one of these stages, it could influence the final recollection.
An important limitation is that the 2024 review focused on adults aged 18 and above. Its findings should not automatically be applied to younger people.
Still, the research raised an interesting question for me. If different cognitive systems influence the way memories are formed and retrieved, could technology help us study these systems together?
That question eventually became the starting point for RECALL.
Introducing RECALL
RECALL is my proposed product concept. It would be a non-invasive brain-computer interface, or BCI, connected to software designed to investigate the processes involved in remembering an event.
The main idea is to look beyond the final answer someone gives when asked what happened.
Instead, RECALL would aim to examine different types of information while someone is recalling an experience. These would include brain activity, their spoken or written recollection, and changes in their physiological responses.
The product would consist of a headset worn on the head, connected to a computer system that collects and analyses the information.
The headset would record electrical activity from the brain. The software would then consider these measurements alongside the person’s recollection and other recorded responses.
I do not imagine RECALL producing a simple result saying that a memory is true or false. That would be an unrealistic claim, especially considering how complicated human memory is.
Get Alia amin’s stories in your inbox
Join Medium for free to get updates from this writer.
Instead, I would like the software to create a visual memory map that organises the information it collects.
For example, it could highlight details that remain consistent across different recollections, details that change, areas where the person expresses uncertainty, and moments that may be worth examining more closely.
It could also compare how confident someone sounds with how consistent their recollection actually is.
Imagine someone describing an event several times. They might remain confident about one particular detail, even though the description changes between recollections. RECALL could highlight that difference for further examination.
It would not automatically decide that the memory is false. It would simply identify something that may deserve more attention.
The hypothesis
The hypothesis behind RECALL is that combining non-invasive neural measurements and physiological responses with a person’s reported recollection could reveal patterns associated with memory distortion that would not be visible from their final answer alone.
The reasoning behind this is that remembering involves several interacting processes. Memory retrieval, attention, source monitoring, confidence, and responses to new information can all influence the way someone recalls an event.
By observing neural activity, reported recollection, and physiological responses together, RECALL could potentially provide researchers with a more detailed understanding of these processes.
The aim is not to identify an objective truth or determine whether someone is being dishonest. It is to investigate whether patterns in these different measurements can help us understand uncertainty, inconsistency, and changes in memory.
The heart-rate sensor
One feature I would particularly like to explore is the addition of a heart-rate sensor.
I initially became interested in the idea of a lie detector, particularly because of the way physical responses can change when someone is under pressure. However, after looking into the science, I realised that a change in heart rate cannot prove that someone is lying.
A person’s heart rate can increase for many reasons, including nervousness, fear, excitement, stress, and concentration. Someone telling the complete truth could still experience these responses, particularly in a stressful situation such as a police interview.
This is why I would not describe the feature as a lie detector. I would rather develop it as a physiological response sensor.
The sensor would monitor changes in a person’s heart rate while they recall an event. These measurements could then be compared with their spoken recollection and the brain activity recorded by the headset.
For example, if a noticeable change in heart rate occurs while someone is discussing a particular detail, the software could mark that moment for further examination.
This might help researchers investigate whether certain parts of a recollection are associated with increased physiological arousal or emotional intensity.
However, the sensor would not be able to identify exactly what someone is feeling or thinking. It would only provide information about physical responses that could have several different explanations.
I think this feature could add another interesting dimension to RECALL because it would allow the system to consider more than brain activity alone.
The three main sources of information would therefore be:
- Neural activity recorded through the non-invasive BCI headset.
- 2. The person’s spoken or written recollection.
- 3. Physiological responses, including changes in heart rate.
The software would consider these sources together rather than treating any one measurement as proof of what happened.
The neuroscience behind RECALL
Although RECALL is still a proposed concept, its foundations are connected to developments that are already taking place in neuroscience.
Brain-computer interface research has made progress in understanding how information can be decoded from neural activity. One area that has received particular attention is speech neuroprostheses.
A 2024 review published in Nature Reviews Neuroscience by researchers including Edward Chang examined developments in systems that use neural activity to help decode attempted speech.
Some of this research has involved people with paralysis and implanted electrodes that record neural signals. These systems have shown promising results, but they are very different from a non-invasive EEG headset.
This distinction matters because it would be inaccurate to suggest that current EEG technology can simply read someone’s memories.
An EEG headset records electrical activity through sensors placed on the scalp. It can provide information about patterns of brain activity, but the signals are complex and do not reveal a clear recording of what someone is thinking.
RECALL would therefore require significant research to establish whether the combination of these measurements could provide useful information about memory retrieval.
I find this particularly interesting because the project brings together several areas that are usually studied separately. Psychology helps explain how memories can change, neuroscience helps us understand the processes involved, and technology raises the possibility of studying those processes in new ways.
Limitations and ethical considerations
There are several challenges that would need to be addressed before RECALL could become a working product.
The first is scientific accuracy. Brain activity does not have one simple meaning. Different people can show different patterns, and the same pattern may occur during different mental activities.
EEG also has limitations. Because it records activity from the scalp, it cannot provide the same level of detail as electrodes placed directly inside the brain. Signals can also be affected by movement and other forms of interference.
The heart-rate sensor has limitations too. A change in heart rate does not tell us whether someone is lying, uncertain, frightened, or simply concentrating. It would be important for the software not to confuse physiological arousal with dishonesty.
There is also the possibility that the AI system could misinterpret the information it receives. If a particular pattern is incorrectly labelled, the software could produce a misleading result. This is especially concerning if the technology is ever considered for use in criminal investigations.
For this reason, RECALL should never be treated as a machine that determines guilt or innocence. Even if future research demonstrates that it can identify useful patterns, those findings would need to be carefully validated before the technology could be considered for any high-stakes setting.
Privacy is another major concern. Brain activity and physiological measurements are personal information. Any future version of RECALL would need clear consent procedures, secure data storage, strict limits on who can access the information, and proper oversight.
There is also a wider ethical question. Just because technology may eventually allow us to collect more information about someone’s brain and physical responses, does that mean we should always be allowed to do so?
I believe this is something that must be considered from the beginning of the project, rather than treated as an issue to deal with later.
The purpose of RECALL should be to improve our understanding of memory, not to invade someone’s privacy or make unsupported judgements about their honesty.
Potential future applications
If future research shows that a system like RECALL can reliably identify meaningful patterns in memory retrieval, it could have applications in several areas.
One possibility is forensic psychology. Researchers could use validated versions of this type of technology to investigate how memories change and which factors may make people more vulnerable to misleading information.
It could also be explored as an additional research tool in criminal investigations. For example, it might help investigators identify parts of a recollection that need further examination.
In the much longer term, there could be questions about whether technology like this has a role in legal settings. However, this would require extensive testing, independent scientific evaluation, and strict ethical safeguards.
It should never replace eyewitness testimony, established investigative procedures, or the judgement of legal professionals.
I think the most important potential contribution of RECALL would be encouraging people to look more carefully at the process behind a memory, rather than accepting the final version of an event without considering how it was formed.
What could RECALL change?
The more I researched this topic, the more I realised that the most interesting part of RECALL is not necessarily the headset or the software.
It is the possibility of asking a different set of questions about memory.
Instead of only asking someone what happened, we could also investigate what they originally noticed, what information they encountered afterwards, how their recollection changed, and whether their confidence remained consistent.
We could also study whether changes in neural activity or physiological responses occur while certain details are being recalled.
None of these measurements would give us a perfect version of the past. They would simply provide different pieces of information that could help researchers understand the process of remembering.
This is what makes the idea interesting to me. It connects something we all experience, remembering, with a much larger issue involving neuroscience, technology, and justice.
Conclusion
What started as an interest in psychology eventually led me to a question I had never really considered before: how much can we actually trust our memories?
The research showed me that memory is not a perfect recording. It can be influenced by the way we perceive an event, the information we encounter afterwards, and the processes involved in retrieving it.
It also showed me that someone can be completely sincere and still remember something incorrectly.
That distinction is especially important when another person’s future could depend on what a witness remembers.
RECALL is my attempt to explore whether technology could help us understand these processes in more detail. By combining neural activity, reported recollection, and physiological responses, the proposed system would investigate different aspects of memory retrieval rather than relying on a person’s final answer alone.
There is still a great deal of research needed before something like this could become a reliable product. It cannot read minds, prove that someone is lying, or determine what objectively happened. I think being honest about these limitations is just as important as being ambitious about the idea itself.
I started this project because I wanted to understand the human mind. Along the way, I found myself exploring neuroscience, artificial intelligence, brain-computer interfaces, and the criminal justice system.
More than anything, this research changed the way I think about memory. We use it every day, often without questioning it. Yet the same memory that helps us understand our own lives can sometimes become an uncertain source of information when another person’s future is involved.
I do not think the answer is to stop trusting our memories. It is to understand them better, recognise their limitations, and keep questioning how they work.
That is what I hope RECALL can contribute to.
References
Innocence Project. (2020). DNA Exonerations in the United States (1989 — 2020). https://innocenceproject.org/dna-exonerations-in-the-united-states/
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. https://doi.org/10.1016/S0022-5371(74)80011-3
Schooler, J. W., & Engstler-Schooler, T. Y. (1990). Verbal overshadowing of visual memories: Some things are better left unsaid. Cognitive Psychology, 22(1), 36 — 71. https://doi.org/10.1016/0010-0285(90)90003-M
Johnson, M. K., Hashtroudi, S., & Lindsay, D. S. (1993). Source monitoring. Psychological Bulletin, 114(1), 3 — 28. https://doi.org/10.1037/0033-2909.114.1.3
Schacter, D. L., Norman, K. A., & Koutstaal, W. (1998). The cognitive neuroscience of constructive memory. Annual Review of Psychology, 49, 289 — 318. https://doi.org/10.1146/annurev.psych.49.1.289
Brassil, M., O’Mahony, C., & Greene, C. M. (2024). Do cognitive abilities reduce eyewitness susceptibility to the misinformation effect? A systematic review. Psychonomic Bulletin & Review, 31, 2410 — 2436. https://doi.org/10.3758/s13423-024-02512-5
National Research Council. (2003). The Polygraph and Lie Detection. The National Academies Press. https://doi.org/10.17226/10420
Silva, A. B., Littlejohn, K. T., Liu, J. R., Moses, D. A., & Chang, E. F. (2024). The speech neuroprosthesis. Nature Reviews Neuroscience, 25, 473 — 492. https://doi.org/10.1038/s41583-024-00819-9
