You’ve heard the saying, “Beauty is in the eye of the beholder,” which captures the idea that beauty is subjective. The same could probably be said about truth, especially when it comes to memory. Perceptions of how accurate a communicated memory is aren’t necessarily shaped by how accurate the memory actually is, but by the linguistic characteristics of the way it’s communicated. In other words, perceived truth may be in the words of the beholder.

This has real consequences, especially in the courtroom.  In a courtroom, the words of eyewitness testimonies can change the outcome of a case. Memories of a crime are communicated to a jury, and the jury is expected to judge the truthfulness of them. But what actually makes a memory sound truthful and accurate to a jury? Is it the amount of detail someone includes? How honest and authentic they seem? Or maybe how logically they tell their story?

In a recent study published in Psychonomic Bulletin & Review, researchers Steven A. Martinez, Kate Cliver, William J. Mitchell, Helen Schmidt, Virginia Ulichney, Chelsea Helion, Jason Chein, and Vishnu Murty strayed away from typical laboratory studies to examine how the linguistic characteristics of communicating threatening memories can affect perceptions of memory accuracy. They creatively chose to explore this question using a haunted house, an environment that naturally introduces a sense of threat.

The authors ran two separate, yet interconnected experiments. Experiment 1 examined how threat and memory characteristics, such as accuracy and level of detail, relate to memory communication styles. Experiment 2 explored how these memory characteristics and communication styles influenced perceptions of accuracy in a new group of participants.

Experiment 1: Sharing a spooky memory

In Experiment 1, participants walked through a haunted house split into different segments that were considered high or low threat. One week after participating in the haunted house, the researchers tested two separate components of their memories of the experience.

First, they examined participants’ episodic specificity by asking them to recall as many details as possible about each separate segment of the haunted house. The researchers focused on the “internal” details in their responses. These internal details are specifically related to events occurring within a specific time or place and are thought to index memory for episodic events.

They also tested the temporal accuracy of their memory, by showing participants pairs of descriptions of events that had occurred in a particular segment of the haunted house and having them choose which one had occurred first within the segment. This allowed the researchers to measure how accurately participants remembered the order of events within the haunted house.

To see how these measures of memory specificity and accuracy relate to communication styles, Martinez and colleagues used a linguistic analysis software called LIWC2022 to analyze the participants’ written descriptions of their recall of the haunted house segments. This linguistic software works by coding the individual words within text into different meaningful categories. They focused on two categories that the LIWC captures: Authenticity and Analytical Thinking.

Authenticity reflects how open or filtered someone sounds when they communicate. A higher score means the person’s language tends to be more spontaneous and natural, like how people talk in everyday conversations. A lower score suggests a more careful or polished style, closer to prepared speech.

Analytical Thinking examines the structure and logic of the language. Someone with a high score tends to sound more organized, formal, and objective. A lower score, on the other hand, usually reflects a more personal and narrative style of speaking.

The main goal of Experiment 1 was to understand how these two different measures of linguistic properties of communicated memories (Authenticity and Analytical Thinking) relate to memory specificity and temporal accuracy.

They found that haunted house memories with greater episodic specificity (more details related to a particular time and place) were expressed with higher levels of Analytic Thinking, but not Authenticity (second column in the figure below).

The opposite pattern was observed for temporal order memory, with more accurate memory of the correct order of events associated with more Authenticity but less Analytic Thinking (third column in the figure below).

They also found that participants used more Authentic and Analytical Thinking when recalling the high-threat parts of the haunted house compared to when they discussed the lower-threat ones (first column in the figure below).

To briefly summarize Experiment 1, Analytical Thinking was linked to detailed memories, and Authentic Language was linked to better memory of event order.

Experiment 2: Judging the spooky memory

Next, Martinez and colleagues wanted to understand whether these four different features of memory communication (Authenticity, Analytic Thinking, episodic specificity, and temporal order memory accuracy) also influence how people judge the accuracy of the memory being shared, much like a jury evaluating an eyewitness account.

To put this to the test, they brought in a new group of participants to play the role of the jury, judging the accuracy of the eyewitness accounts from the original haunted house storytellers.  Across multiple trials, they showed these participants two memory transcripts side by side. Each of the two transcripts described the same segment from the haunted house, but they each came from two different people in Experiment 1. Their task was to simply read both transcripts and select which one seemed more accurate. You can see an example of this setup in the figure below.

They found that greater use of Authentic language when recalling the haunted house segments did not increase the likelihood that the transcript of their memory would be judged as more accurate than the other transcript option by the recipient of the communicated memory (see left graph below).

Interestingly, higher levels of Analytic Thinking did increase the chances that a transcript was perceived as more accurate (right graph). In other words, Authenticity didn’t sway accuracy judgements, but Analytic Thinking did, making people more likely to believe the memory was accurate.

In the first experiment, the researchers used temporal order memory as their only direct measure of accuracy. Even though this was a concrete and objective measure, they found in Experiment 2, that it didn’t actually match up with how people judged accuracy (left graph below). Instead, what really made a difference was episodic specificity. The more detailed the memory, the more accurate it seemed to others (right graph below).

To summarize, people tended to rely on an Analytic Thinking linguistic style and a high level of episodic details as cues for accuracy, even though these features don’t really reflect how accurate a memory truly is. At the same time, they tend to undervalue Authenticity language, even though higher Authenticity was actually linked to higher accuracy, specifically in terms of remembering the correct order of events.

What do these findings mean for courtroom testimonies? According to the featured study, people are more likely to believe in detailed memories, not necessarily those that are authentic. That is, truth in memory isn’t just about what occurred, it’s about how a story is told.

Featured Psychonomic Society article

Martinez, S. A., Cliver, K., Mitchell, W. J., Schmidt, H., Ulichney, V., Helion, C., … & Murty, V. (2025). Linguistic properties of memory expression differentially relate to accuracy, specificity, and perceived veracity. Psychonomic Bulletin & Review, 1-11. https://doi.org/10.3758/s13423-025-02667-9