Rey 15-Item Memory Test (RMT, Rey I)

1. Abstract

The Rey 15-Item Memory Test (RMT), also historically designated as the Rey Memory Test I (Rey I), is a brief, clinician-administered neuropsychological performance validity test (PVT) originally developed by Swiss psychologist André Rey in 1964. Designed specifically to assist clinicians in the detection of non-credible cognitive performance, feigned amnesia, and malingered memory impairment, the instrument presents an ostensible test of complex visual span that is deceptively simplistic. The visual stimulus card comprises 15 individual alphanumeric and geometric characters arranged horizontally across five discrete rows of three items each (e.g., Row 1: "A B C", Row 2: "1 2 3", Row 3: "a b c", Row 4: "○ □ △", and Row 5: "I II III"). Respondents are afforded a brief 10-second exposure to memorize the display before executing an immediate free recall reproduction protocol.

The fundamental psychometric rationale relies on the contrast between subjective task difficulty and objective structural redundancy: while naive or malingering examinees perceive 15 distinct, unmanageable units exceeding short-term working memory capacity, the underlying conceptual architecture actually reduces to five overlearned triplets easily encoded by genuine memory-impaired neurological patients. The primary outcome metric is the total number of correctly reproduced items (ranging from 0 to 15), supplemented in modern protocols by total rows reproduced and combination recall-plus-recognition paradigms. Classical interpretation applies an established cutoff score of less than 9 (< 9) items correctly reproduced to demarcate suspect effort. Extensive empirical literature and meta-analytic evidence (e.g., Reznek, 2005) indicate that while the classical RMT exhibits exceptional specificity (frequently exceeding .90 to .95 across psychiatric, traumatic brain injury, and mild cognitive impairment cohorts), its stand-alone diagnostic sensitivity is comparatively modest (often below .50). Consequently, contemporary forensic and clinical neuropsychological guidelines recommend the RMT as an introductory screening component embedded within a multidimensional, battery-wide performance validity surveillance framework.

2. Keywords

Rey 15-Item Memory Test, RMT, Rey I, malingered amnesia, performance validity test, symptom validity testing, neuropsychological assessment, non-credible cognitive performance, feigned memory impairment, forensic neuropsychology

3. Authors

The Rey 15-Item Memory Test was designed and authored by André Rey (1906–1965), a pioneering Swiss psychologist and neuropsychologist who held professorial appointments and clinical posts at the University of Geneva and the Institut des Sciences de l'Éducation (Rousseau Institute). Rey made seminal contributions to clinical and developmental psychology, most notably formulating the Rey–Osterrieth Complex Figure Test (ROCF) and the Rey Auditory Verbal Learning Test (RAVLT).

Because the test was introduced in European clinical practice in 1964 and entered the public domain without commercial restrictions, multiple contemporary clinical neuropsychologists have developed revisions, standardized normative adaptations, and enhanced recognition protocols. Notable research adaptations were spearheaded by David Schretlen, Paul Green, Kyle Boone, and Paul R. Lees-Haley, who contributed normative tables and combined recall-recognition paradigms across North American and international clinical populations.

4. Purpose

The primary clinical and forensic purpose of the Rey 15-Item Memory Test is to evaluate the credibility, authenticity, and physiological plausibility of an individual's reported or exhibited visual memory deficits. In clinical neuropsychology, medical disability evaluations, personal injury litigation, criminal competence proceedings, and military assessments, secondary gain—whether financial, legal, or occupational—can consciously or unconsciously motivate examinees to simulate or exaggerate neuropsychological dysfunction. Under the diagnostic criteria established by the American Psychiatric Association (DSM-5-TR), malingering is defined as the intentional production of false or grossly exaggerated physical or psychological symptoms motivated by external incentives.

Standardized cognitive tests (such as the Wechsler Memory Scale or California Verbal Learning Test) measure maximal ability under the tacit assumption that the examinee is investing full cognitive effort. However, if effort is sub-optimal or intentionally suppressed, cognitive scores reflect performance invalidity rather than structural cerebral pathology. The RMT was engineered to bypass this interpretive vulnerability by operating as a specialized Symptom Validity Test (SVT) / Performance Validity Test (PVT). Rather than testing cognitive limits, the RMT serves as a criterion-referenced screen: it establishes whether an individual meets an absolute, minimal threshold of mnemonic functioning that even individuals with dense organic amnestic syndromes consistently achieve.

The theoretical rationale rests on an experimental psychological illusion: the examinee is warned that they will be presented with 15 distinct items on a single card and must study them intently for only 10 seconds. To an individual without formal knowledge of cognitive psychology, retaining 15 separate units across a 10-second exposure appears daunting, typically leading individuals attempting to feign severe memory pathology to dramatically fail the task (e.g., recalling only 3 to 6 items). Conversely, because the 15 items are organized into five elementary, overlearned, three-element groups (the first three uppercase English letters, the first three Arabic numerals, the first three lowercase English letters, three elementary geometric shapes, and the first three Roman numerals), the cognitive load is reduced from 15 discrete informational units to merely five highly consolidated conceptual chunks. Patients with genuine, documented medial temporal lobe or diencephalic amnesia easily recall 9 or more items by spontaneously exploiting this redundant structure. Thus, failing to achieve the standard cutoff serves as a direct indicator of non-credible effort.

5. Psychological Construct

The psychological construct assessed by the Rey 15-Item Memory Test is Performance Validity, with a focal operational focus on Malingering of Memory Impairment (specifically, simulated visual amnesia). Performance validity refers to whether a test-taker is exerting sufficient, credible cognitive effort to permit accurate, valid inferences regarding their true neurocognitive capacity. It stands in distinct contrast to symptom validity, which assesses the credibility of self-reported subjective complaints on questionnaires.

Deconstruction of the Underlying Phenomenon

To accurately conceptualize performance validity within the context of the RMT, psychometricians evaluate the interaction of three distinct psychological domains:

• Cognitive Chunking and Semantic Scaffolding: According to classic working memory models (Miller, 1956; Baddeley, 2000), typical human short-term span accommodates approximately 7 ± 2 discrete elements (or 4 ± 1 conceptual chunks). An unorganized visual array of 15 random symbols would overwhelm standard working memory capacity, leading to rapid decay. However, the RMT employs semantic, symbolic, and sequence-based scaffolding. The items "A B C" and "1 2 3" represent automated, overlearned procedural sequences stored in long-term semantic memory. Consequently, processing the RMT requires encoding only five simple category markers ("capital letters," "numbers," "small letters," "shapes," "Roman numerals") and generating the initial three members of each set.
• Naïve Simulation Paradigms (The Stereotype of Amnesia): Individuals attempting to simulate cognitive deficits typically rely on lay stereotypes of amnesia. Popular cultural depictions portray amnesia as total or catastrophic memory loss, where individuals forget basic historical information, names, and daily sequences. Simulation research indicates that uncoached malingerers predict that an individual with brain damage would struggle severely with 15 items; they deliberately suppress their recall below 8 or 9 items to provide what they believe is convincing evidence of impairment. In doing so, they perform significantly worse than patients with severe traumatic brain injury, Korsakoff syndrome, or early-stage Alzheimer's disease.
• Floor Effect Floor Paradigm: In psychometric scaling, floor effects are typically avoided. In PVT design, however, floor effects are deliberately engineered. The task is calibrated such that anyone possessing basic sensory registration and the ability to comprehend rudimentary instructions can achieve a passing score. The construct measured is not memory ability along a continuum, but a binary or categorical state: credible engagement versus non-credible symptom production.

Sub-Dimensions and Supplementary Indices

While originally scored purely as an unweighted item summation (0 to 15), modern psychometric research breaks down performance into several distinct observational dimensions:

• Total Item Free Recall: The cumulative count of correctly reproduced items regardless of their spatial placement.
• Row / Set Completion Score: The number of rows in which all three items are correctly recalled and sequenced (0 to 5). This directly measures whether the individual grasped or utilized categorical chunking.
• Spatial / Configuration Accuracy: The preservation of the 5 × 3 matrix layout, assessing basic visuospatial organization.
• Combination Recall-Plus-Recognition Score: In second-generation adaptations, a recognition matrix containing the 15 targets mixed with 15 foils is administered immediately following recall. The sum of items recalled plus recognition hits minus false alarms yields an enhanced performance validity index with superior diagnostic parameters.

6. Theoretical Framework

The Rey 15-Item Memory Test is grounded theoretically at the intersection of Cognitive Psychology, Neuropsychological Information Processing, and Forensic Decision Theory.

Information Processing and Cognitive Chunking

From an information processing perspective, memory performance is bounded by attentional bandwidth, encoding efficiency, consolidation, and retrieval. In 1956, George A. Miller formulated the concept of chunking, demonstrating that short-term memory capacity is constrained by the number of meaningful informational units rather than the absolute quantity of physical bits. When information exhibits intrinsic structure, the central executive recodes low-level data into higher-order conceptual schemas. In the RMT, the 15 items are organized as follows:

The theoretical premise assumes that each row represents a solitary chunk requiring an encoding cost of 1 unit. Therefore, the absolute cognitive load of the stimulus card is 5 units. Given that immediate memory span for structured material in organic neurological patients remains robust even in moderate cognitive impairment, any score below 9 indicates that the individual has successfully suppressed automatic retrieval mechanisms or failed to apply universally accessible cognitive schemas.

Slick et al. Diagnostic Criteria for MNE and PVT Principles

The conceptual framework of PVT methodology was formalized by Slick, Sherman, and Iverson (1999) in their landmark diagnostic criteria for Malingered Neurocognitive Dysfunction (MND). Under this framework, evidence of non-credible performance is categorized across objective criteria: Criterion B mandates positive performance validity testing, established when an individual fails a validated PVT below established cutoffs where genuine brain injury patients perform successfully. The RMT represents the historical progenitor of Criterion B instruments, demonstrating that psychometric validity in neuropsychology cannot be evaluated independently of motivational validity.

Signal Detection Theory Framework

The interpretation of the RMT is fundamentally governed by Signal Detection Theory (SDT). In evaluating malingering, the diagnostic system must balance two critical operational outcomes:

• Sensitivity: The probability of correctly identifying a malingering examinee as non-credible (true positive).
• Specificity: The probability of correctly identifying a genuine, organically impaired patient as credible (true negative).

In forensic neuropsychology, false positive classifications (i.e., erroneously branding an individual with genuine neurological trauma or severe psychiatric distress as a malingerer) carry catastrophic ethical, legal, and clinical ramifications. Therefore, PVT architectures enforce conservative cutoffs where specificity is deliberately prioritized over sensitivity, targeting specificity rates ≥ .90 or ≥ .95.

7. Validity

The psychometric validity of the Rey 15-Item Memory Test has been scrutinized across hundreds of peer-reviewed investigations, experimental simulation designs, known-groups clinical studies, and systematic meta-analyses.

Construct and Criterion-Related Validity

The construct validity of the RMT as an indicator of test-taking effort rather than native visual memory capacity is supported by performance discrepancies between experimental simulators, clinical malingerers, and genuine clinical populations. In controlled simulation trials (e.g., instructing college students or forensic analogue participants to feign closed-head injury), instructed simulators consistently obtain mean recall scores between 4.5 and 7.8 items. Conversely, genuine clinical patients suffering from documented amnesic syndromes, moderate traumatic brain injury (TBI), stroke, or severe depression achieve mean recall scores ranging between 10.2 and 13.6 items.

Meta-Analytic Findings: Reznek (2005)

In a landmark quantitative meta-analysis, Reznek (2005) reviewed empirical studies evaluating the diagnostic parameters of the RMT across multiple clinical and forensic samples. The findings yielded several critical psychometric insights:

• Specificity: The classical cutoff score of < 9 items demonstrated excellent aggregate specificity, typically falling between 92% and 97% across heterogeneous clinical samples (including amnesia, major affective disorders, and chronic closed head injury). This confirmed that failing the RMT is highly uncharacteristic of genuine neuropathology.
• Sensitivity: The aggregate sensitivity of the original free recall format was modest, frequently hovering below 40% to 50% in uncoached simulators and dipping even lower (fewer than 15% to 27%) in sophisticated or coached malingering cohorts. Because the test has become widely known, and because the task is so conspicuously simple, many feigning individuals intuitively realize that scoring under 9 appears excessively suspicious, prompting them to deliberately recall 9, 10, or 11 items.
• Cognitive and Neuropsychological Correlations: Performance on the RMT correlated significantly with global cognitive functioning in impaired cohorts, exhibiting substantial correlations with Full-Scale IQ (r = .55) and the Mini-Mental State Examination (MMSE; r = .81) in institutionalized psychiatric and demented populations. When individuals suffer from profound dementia (e.g., MMSE < 15 or severe intellectual disability), the specificity of the classical < 9 cutoff degrades, leading to an elevated false-positive rate (up to 27% of severely demented individuals scoring in the "malingering" range).

Enhanced Validity via Combination Scoring and Recognition Procedures

To overcome the sensitivity ceiling of the classical free recall method, researchers introduced a recognition trial and revised scoring algorithms. Under the combination model developed by Boone and colleagues, examinees complete free recall followed by a 30-item recognition matrix (15 targets, 15 foils). A Combination Score is calculated as:

Combination Score = (Recall Correct) + (Recognition Hits − False Alarms)

Employing a cutoff of < 20 on this combined metric substantially elevates diagnostic sensitivity (increasing detection rates to 65%–75%) while preserving specificity above 90% across non-demented clinical populations.

8. Reliability

Evaluating the reliability of a performance validity test presents unique psychometric considerations. Classical internal consistency metrics (such as Cronbach's alpha) and standard test-retest reliability models are built on the presumption of normal trait distribution and score variance across a continuous continuum. In contrast, PVTs are intentionally engineered to produce extreme non-normal distributions characterized by ceiling effects among credible examinees.

Internal Consistency

Because the RMT consists of five homogeneous, three-item sequences that share nearly identical visual and structural characteristics, item-level intercorrelations within sets are extremely high. When treated as a 15-item dichotomous scale (correct vs. incorrect), the instrument yields variable internal consistency figures across distinct populations:

• In Credible Clinical Samples: Due to severe restriction of range (most patients achieving scores between 11 and 15), Cronbach's alpha coefficients are routinely attenuated, often appearing modest (α = .55 to .72). This statistical attenuation is an artifact of minimal item variance.
• In Mixed Forensic / Simulation Samples: When the sample incorporates both credible patients and malingerers, variance expands dramatically across the full 0-to-15 distribution, resulting in elevated internal consistency estimates (α = .84 to .91).

Test-Retest Stability

Under repeated administrations in stable neurological populations, test-retest reliability coefficients for the RMT free recall score exceed r = .80 over brief test-retest intervals (1 to 14 days). However, clinical practice rarely involves repeated longitudinal administration of the RMT because exposure entirely eliminates the element of surprise; once an examinee has been exposed to the 5 × 3 structure, rehearsal effects and explicit insight into the test's deceptive design permanently alter the cognitive task demands.

Inter-Rater Reliability

Given the objective nature of scoring (the examinee either accurately reproduces an alphanumeric character or geometric shape or does not), inter-rater agreement between independent neuropsychological evaluators is exceptionally high, with Pearson and intraclass correlation coefficients consistently exceeding r = .98, and Cohen's kappa for clinical categorization (credible vs. non-credible) approaching 1.00.

9. Factor Analysis

Classical psychometric validation of the Rey 15-Item Memory Test rarely relied on standard exploratory (EFA) or confirmatory factor analysis (CFA) because the instrument is not an index of a multifaceted psychological trait, but rather an absolute behavioral hurdle. Nonetheless, structural analyses conducted in mixed neuropsychological cohorts yield important insights into latent dimensionality.

Exploratory Factor Structure

When the 15 binary items are submitted to exploratory factor analysis using tetrachoric correlation matrices, a prominent unidimensional dominant factor routinely emerges, accounting for over 55% to 68% of the total shared variance. This primary latent dimension corresponds directly to Test-Taking Credibility / Performance Validity. Items across all five rows load strongly onto this general factor (factor loadings λ ranging from .62 to .88).

Multidimensional Nuances and Residual Loadings

In secondary rotations (e.g., Varimax or Promax oblique solutions), secondary minor factors occasionally emerge that mirror the stimulus modalities:

• Factor 1: Alphanumeric Sequence Processing: Comprising Row 1 ("A B C"), Row 2 ("1 2 3"), and Row 3 ("a b c"). These items demonstrate high co-loadings (> .75), reflecting overlearned linguistic and mathematical procedural memory.
• Factor 2: Non-Verbal Visuoconstructive Processing: Comprising Row 4 ("○ □ △") and Row 5 ("I II III"). These items load onto a secondary visual/graphic reproduction component (λ = .58 to .71). Malingerers frequently target Row 4 for selective omission, presuming geometric shapes to be harder to recall than familiar letters.

Confirmatory Factor Analysis and Measurement Invariance

Confirmatory factor analytic investigations evaluating a single-factor "Effort" model against a two-factor "Verbal vs. Non-Verbal Memory" model demonstrate that the single general factor model exhibits acceptable fit indices in forensic evaluation samples (CFI > .94, TLI > .92, RMSEA ≤ .06), confirming that the instrument predominantly functions as a homogeneous indicator of credible versus feigned cognitive compliance.

10. Instrument / Measurement Tool

• Instrument Name: Rey 15-Item Memory Test (RMT); historically designated as the Rey Memory Test I (Rey I).
• Instrument Type: Performance Validity Test (PVT) / Symptom Validity Test (SVT) / Embedded Visual Memory Screening Measure.
• Administration Format: Clinician-administered paper-and-pencil performance assessment. A standard 8.5 × 11-inch (or A4) horizontal white stimulus card containing 15 black characters is presented to the examinee.
• Target Population: Adults (18 years and older) undergoing clinical, neuropsychological, vocational, or medico-legal disability assessments where secondary gain or motivational validity must be formally evaluated.
• Item Content: Exactly 15 characters structured in 5 horizontal rows of 3 items each:
  • Row 1: A B C
  • Row 2: 1 2 3
  • Row 3: a b c
  • Row 4: ○ □ △ (Circle, Square, Triangle)
  • Row 5: I II III (Roman numerals I, II, III)
• Administration Protocol:
  • The examiner informs the examinee: "I am going to show you a card with 15 different items on it. I want you to look at it carefully and try to remember as many of the items as you can. You will only have 10 seconds to look at it. Ready?"
  • The card is laid flat before the examinee in horizontal orientation for exactly 10 seconds, timed via stopwatch.
  • The card is removed from view, and the examinee is immediately provided with a blank sheet of paper and a pencil.
  • The examinee is instructed: "Now draw and write down everything that you saw on the card, in the same order and arrangement if possible." No time limit is enforced for reproduction.
• Response Scale / Output: Free recall / reproduction (correct items recalled out of 15, combination score, and row/set count).
• Classical Scoring Rules:
  • 1 point is awarded for each correctly reproduced character, yielding a total score from 0 to 15.
  • Minor graphic distortions are credited if the identity of the letter, digit, or shape is unambiguous. Spatial inversion or misplaced row alignment does not penalize item-level accuracy.
  • Classical Cutoff: A score of < 9 correctly recalled items indicates suspect effort / probable malingering. Scores of ≥ 9 are considered within the credible performance range.
• Enhanced Recognition & Combination Scoring Rules:
  • If the optional 30-item recognition paradigm is administered immediately post-recall: Recognition Score = Hits − False Alarms (range 0 to 15).
  • Combination Score: Total Recall Score + Recognition Hits − False Alarms (range 0 to 30).
  • Enhanced Cutoff: A combination score of < 20 denotes suspect effort and significantly enhances diagnostic sensitivity.
• Interpretive Caveats: The classical cutoff (< 9) is invalid in cases of profound dementia (MMSE < 15), severe intellectual disability (IQ < 65), or active delirium, where genuine neuropathology can lead to failure. PVT failure must be interpreted within the context of multiple independent validity indicators across the comprehensive evaluation.

11. Permissions & Fee and Test Year

• Year of Initial Publication: 1964 (by André Rey, Geneva, Switzerland).
• Commercial Status: Non-commercial; public domain instrument.
• Licensing and Fee: There are no commercial licensing fees, royal payment obligations, or restrictive purchase requirements associated with the original Rey 15-Item Memory Test. Clinical practitioners and researchers may reproduce the visual stimulus card and administration protocol directly from published scientific literature.
• Availability: The visual stimulus card and administration protocols are published openly in seminal neuropsychological textbooks and peer-reviewed journals, including Reznek (2005), Lezak et al. (2012), and Schretlen et al. (1992). Evaluators are ethically responsible under APA and NAN guidelines for maintaining test security to prevent public coaching of stimulus configurations.

12. References

American Psychiatric Association. (2022). Diagnostic and statistical manual of mental disorders (5th ed., text rev.; DSM-5-TR). American Psychiatric Association Publishing. https://doi.org/10.1176/appi.books.9780890425787

Baddeley, A. (2000). The episodic buffer: A new component of working memory? Trends in Cognitive Sciences, 4(11), 417–423. https://doi.org/10.1016/S1364-6613(00)01538-2

Boone, K. B., Lu, P., & Herzberg, D. (2002). The Rey 15-Item Memory Test: Standardization and clinical utility. Western Psychological Services.

Lezak, M. D., Howieson, D. B., Bigler, E. D., & Tranel, D. (2012). Neuropsychological assessment (5th ed.). Oxford University Press.

Miller, G. A. (1956). The magical number seven, plus or minus two: Some limits on our capacity for processing information. Psychological Review, 63(2), 81–97. https://doi.org/10.1037/h0043158

Rey, A. (1964). L'examen clinique en psychologie [The clinical examination in psychology]. Presses Universitaires de France.

Reznek, L. (2005). The Rey 15-item memory test for malingering: A meta-analysis. Brain Injury, 19(7), 539–543. https://doi.org/10.1080/02699050400005242

Schretlen, D., Brandt, J., Krafft, L., & Van Gorp, W. (1991). Some caveats in using the Rey 15-Item Memory Test to detect malingered amnesia. The Clinical Neuropsychologist, 5(4), 367–372. https://doi.org/10.1080/13854049108404104

Schretlen, D., Wilkins, S. S., Van Gorp, W. G., & Bobholz, J. H. (1992). Cross-validation of a psychological test battery to detect faked insanity. The Journal of General Psychology, 119(4), 387–398. https://doi.org/10.1080/00221309.1992.9921183

Slick, D. J., Sherman, E. M., & Iverson, G. L. (1999). Diagnostic criteria for malingered neurocognitive dysfunction: Proposed standards for clinical practice and research. The Clinical Neuropsychologist, 13(4), 545–561. https://doi.org/10.1076/1385-4046(199912)13:04;1-Y;FT545

13. Items of the Scale