What Is Reality Monitoring in Deception Research?

By Murder Book Editorial TeamPublished by Gungrounds, the studio making Murder BookPublished 3 October 2026Updated 3 October 2026

Reality monitoring (RM) is a theory about how people tell the difference between things they actually experienced and things they only imagined, dreamed or were told about. Psychologists Marcia Johnson and Carol Raye proposed the framework in 1981, arguing that real memories tend to contain more sensory detail, spatial information and emotional content, while imagined memories contain more references to cognitive effort. Deception researchers later adapted the idea: if genuine memories have distinctive qualities, then a truthful statement about a real event should show those qualities more strongly than a fabricated one. Studies have found that RM criteria can distinguish truth from lies at rates above chance, but accuracy is far from certain and the method has important limits.

Key takeaways

Reality monitoring is based on Johnson and Raye's 1981 theory that memories of real events differ in quality from memories of imagined events.
Vrij proposed eight RM criteria for deception detection, including clarity, perceptual information, spatial information and cognitive operations.
Research suggests RM can reach accuracy rates of around 70%, similar to criteria-based content analysis, but individual-level judgments remain uncertain.
Truthful statements tend to score higher on sensory and contextual criteria; fabricated statements tend to score higher on cognitive operations.
No single verbal analysis method reliably determines whether a specific person is lying.

Where does reality monitoring come from?

Marcia Johnson and Carol Raye published their reality monitoring framework in Psychological Review in 1981. The core question was simple: how do people know whether a memory reflects something that actually happened or something they only thought about?

Their answer was that external memories (those based on real perceptions) and internal memories (those generated by thought, imagination or dreams) differ systematically in their qualities. External memories tend to be richer in sensory details (what something looked like, sounded like, felt like), spatial information (where things were) and temporal information (when things happened). Internal memories tend to contain more traces of cognitive operations, the mental effort involved in generating or reasoning about the memory.

This was originally a theory about memory, not about deception. But the logic invited an obvious application: if someone describes an event they actually experienced, their account should carry the hallmarks of an external memory. If they are making something up, their account should look more like an internal one.

How has reality monitoring been applied to deception detection?

Aldert Vrij adapted the RM framework into a set of criteria that could be used to assess whether a verbal statement is likely truthful or fabricated. Vrij proposed eight criteria: clarity, perceptual information, spatial information, temporal information, affective information, reconstructability (whether the events described can be reconstructed from the account), realism and cognitive operations.

The first seven criteria are expected to appear more in truthful statements. Cognitive operations, which reflect the thinking process behind a statement ("I must have..." or "I think it was..."), are expected to appear more in fabricated ones.

In practice, a trained coder reads or listens to a statement and rates it on each criterion. The pattern of scores then suggests whether the account is more consistent with a real experience or an imagined one.

What does the research show?

A 2005 review by Masip, Sporer, Garrido and Herrero in Psychology, Crime and Law examined the empirical evidence for the RM approach to deception detection. They found that few individual criteria discriminated consistently across studies, and several variables moderated the effects. However, the approach as a whole discriminated above chance, reaching accuracy rates similar to those of criteria-based content analysis.

A large meta-analysis by Oberlader and colleagues, published in Law and Human Behavior in 2016, pooled 55 studies testing both CBCA and RM. It found a large overall effect size (g = 0.98) for verbal content analysis methods, with no significant difference in the effectiveness of CBCA and RM. This suggests that both approaches tap into genuine differences between experienced and fabricated accounts, though neither achieves the level of accuracy needed for confident individual judgments.

Research has found that the criteria most consistently linked to truthfulness include perceptual details, spatial details and contextual embedding. Cognitive operations, as predicted, tend to appear more in fabricated statements, though this finding is less consistent across studies.

How does reality monitoring differ from CBCA?

Both RM and CBCA analyse the content of a verbal statement to assess credibility, but they come from different theoretical traditions and focus on different features.

CBCA grew out of forensic practice in German-speaking countries and was designed to assess children's testimony in sexual abuse cases. It uses 19 criteria grouped into five categories, including features like unexpected complications, unusual details and self-deprecation. It is more complex and more subjective in its application.

RM is grounded in cognitive psychology and memory theory. Its criteria are fewer, more directly tied to memory science, and arguably easier to apply consistently. A 2005 review by Vrij noted that CBCA evaluations were not accurate enough for courtroom expert testimony but might be useful in investigations. The same caution applies to RM.

Both methods also differ from the verifiability approach, which focuses not on whether a statement sounds like a real memory but on whether its details could be checked by an investigator.

What are the limits of reality monitoring for lie detection?

Several important caveats apply.

The theory was designed for memory, not deception. Johnson and Raye's framework describes how people distinguish their own memories. Applying it to detect other people's lies is an extrapolation. A liar who incorporates real sensory details from genuine experiences into a fabricated account may produce a statement that scores high on RM criteria.

Individual accuracy is uncertain. Group-level differences do not mean that any single statement can be reliably classified. An honest person giving a vague account of a stressful event may score lower on sensory detail than a prepared liar who rehearses vivid descriptions. Research on why people say "I don't remember" during questioning shows that genuine memory gaps are common.

Cultural and individual differences matter. People vary in how much sensory detail they naturally include in their accounts. Differences in narrative style, education, language fluency and emotional state can all affect RM scores independently of whether the person is telling the truth.

Coding requires training. Although RM criteria are simpler than CBCA criteria, applying them consistently still requires judgment. Studies have found acceptable inter-rater reliability in controlled settings, but real-world application introduces more variability.

No behaviour or verbal pattern reliably proves deception. Meta-analyses, including Bond and DePaulo's 2006 review, consistently show that people judge lies and truths correctly only about 54% of the time. Verbal analysis tools improve on that baseline, but they do not eliminate uncertainty. For more on this point, see Can Body Language Tell You If Someone Is Lying?.

How does Murder Book connect to reality monitoring?

Murder Book: The Art of Interrogation is an upcoming narrative interrogation game from Gungrounds, an independent studio in Zagreb, Croatia. Development screenshots show an in-game casebook listing behaviours such as Fabrication, Emotional Fading and Acting Emotion. The Acting Emotion entry links failing to sustain a performed emotion to cognitive load theory, a concept closely related to the cognitive operations that reality monitoring tracks.

When a suspect in Murder Book tells a story that lacks sensory grounding or leans heavily on reasoning and justification, the player faces the same question that RM researchers study: does this sound like a memory of something that happened, or something that was constructed? The game gives players tools to probe further, not to render a verdict from a single cue.

Murder Book is in development for PC (Steam) and mobile, with release currently planned for the end of Q2 2027.