Recognition Subtests of the Repeatable Battery for the Assessment of Neuropsychological Status: Evidence for a Cortical vs. Subcortical Distinction

Abstract

Introduction:

Within clinical neuropsychology, a classic diagnostic distinction is made between cortical and subcortical disorders, especially based on their memory profiles. Typically, this is based on the comparison of recall and recognition trials, where individuals with cortical conditions do not tend to benefit (i.e., score well) on recognition trials and individuals with subcortical conditions do. Although the Repeatable Battery for the Assessment of Neuropsychological Status (RBANS) is a widely used brief cognitive battery, there is a lack of evidence to support this measure’s utility in distinguishing between the memory profiles of these conditions.

Method:

Thirty-six mild Alzheimer’s disease (AD), 55 Parkinson’s disease (PD), and 105 essential tremor (ET) participants (N = 196) were administered the RBANS with additional Story and Figure Recognition subtests. Group differences on recall and recognition scores (Total Correct, Hits or True Positives, False Positive Errors, and discriminability index) were examined across the three groups, while controlling for the influence of age and gender.

Results:

As expected, individuals with AD had poorer recognition scores compared to the other clinical groups across tasks (all p-values < .05), while the ET sample largely performed comparably to the PD sample. With the exception of comparable Figure Recognition and Recall in the PD sample, all groups exhibited significantly greater recognition Hit performance compared to Recall (all p-values < .05).

Conclusions:

The group differences in performance across RBANS recognition subtests suggest support for traditional “cortical” and “subcortical” profiles. However, all groups, including the mild AD sample, demonstrated a benefit from recognition cues compared to free recall. Overall, these findings support the inclusion of the newly developed Story and Figure Recognition subtests in future clinical practice and research endeavors.

Neuropsychological assessments are important for identifying and characterizing memory impairment to aid in differential diagnosis. A classic but not definite differentiation contrasts “cortical” and “subcortical” dementia profiles (Arango-Lasprilla et al., 2006; Bonelli & Cummings, 2008; Salmon & Filoteo, 2007). For memory tests, a deficit in recognition is a hallmark of cortical profiles, such as for Alzheimer’s disease (AD), and is primarily driven by higher false positive errors, which are thought to reflect difficulties discriminating previously presented information from distractors (Brandt et al., 1992; Clark et al., 2012; Euler et al., 2022; Harciarek & Jodzio, 2005; Hildebrandt et al., 2009). In contrast, subcortical profiles, typically noted for Parkinson’s disease (PD) and Huntington’s disease, are associated with relatively intact recognition despite difficulties with retrieval (Brandt et al., 1992; Salmon & Bondi, 2009; Salmon & Filoteo, 2007). Taken together, differences between free recall and recognition performances are essential considerations for diagnostic impressions of memory impairments.

Notably, there is mixed support for clear neuropathological and cognitive profile distinctions between cortical and subcortical dementias (Arango-Lasprilla et al., 2006; Bonelli & Cummings, 2008). Meta-analytic evidence suggests unignorable performance overlap between “cortical” AD and “subcortical” vascular dementia across commonly used measures (Mathias & Burke, 2009). Indeed, small to medium effect sizes were found for verbal (i.e., word lists; Average weighted effect size = −0.4) and visual (Average weighted effect size = −0.4) recognition memory tests, with a 73% overlap in performances between groups (Mathias & Burke, 2009). Additionally, recognition measures appear to be sensitive to AD, but not specific when compared to subcortical dementias (Weissberger et al., 2017). However, PD appears to be primarily associated with impaired recognition memory for those with dementia (d = 1.30) rather than for those without (d = 0.16), based on a meta-analysis (Whittington et al., 2000). Thus, it is necessary to understand the utility of specific recognition measures in differentiating cortical and subcortical profiles.

The Repeatable Battery for the Assessment of Neuropsychological Status (RBANS) is a frequently used brief test battery that assesses clinically relevant domains of cognitive functioning (Randolph, 1998). This measure reliably characterizes deficits in various clinical populations, and most relevantly provides distinct subcortical and cortical profiles (Beatty et al., 2003; Randolph, 1998). More specifically, the subcortical presentations typical of PD and Huntington’s disease are associated with poorer performance on measures comprising the Visuospatial/Constructional and Attention Indexes on the RBANS (Beatty et al., 2003; Duff et al., 2010; Randolph, 1998). On the other hand, lower scores on Language and Delayed Memory Indexes reflect a cortical profile seen in individuals with AD (Randolph et al., 1998). At the subtest level, the expected differences between classically cortical and subcortical disorders on tests of delayed recall and recognition are present (Duff et al., 2008; Duff et al., 2010; Euler et al., 2022; McDermott & DeFilippis, 2010; Randolph et al., 1998). Notably, the Delayed Memory Index consists of three free recall tasks (i.e., List Recall, Story Recall and Figure Recall); however, until recently, it only had a recognition subtest for the list of words. Duff and colleagues (2021) developed comparable Story Recognition and Figure Recognition subtests, compared samples of amnestic Mild Cognitive Impairment (MCI), mild AD, and cognitively-intact older adults on these new subtests, and found that the impaired groups performed worse than intact individuals on the number of hits, number of false positive errors, and the total score on these recognition subtests. Further, poorer recall and total recognition scores showed comparable associations with biomarkers of AD, including greater global amyloid plaque distribution, reduced bilateral hippocampal volume, and APOE E4 status (Euler et al., 2022). The inclusion of supplemental memory measures has also demonstrated more general clinical usefulness by allowing for the generation of nuanced indices, such as Verbal Memory, Visual Memory, and Recognition/Cueing (Gradwohl et al., 2022). Considered together, these results support the utility of these novel recognition subtests in the context of identifying individuals at risk for AD and improving clinical interpretations of memory functioning. However, to our knowledge, no study has compared performance on these subtests in cortical and subcortical disorders.

Thus, our study aimed to characterize recognition performance on the RBANS (List Recognition, Story Recognition, and Figure Recognition subtests), in samples of individuals diagnosed with mild AD, PD, or essential tremor (ET). Whereas AD reflects a cortically-based condition and PD reflects a subcortical one, the subjects with ET were included as a comparison group that has not clearly demonstrated either such profile, even though they can present with some mild cognitive difficulties (Campbell et al., 2022; Vehar et al., 2023). More specifically, although individuals with ET tend to perform similarly to individuals with PD (Lacritz et al., 2002; Lombardi et al., 2001; Puertas-Martín et al., 2016), different subtypes based on cognition have been proposed to account for within-group heterogeneity (Bhatia et al., 2018; Ratajska et al., 2022). It was hypothesized that (1) individuals with mild AD would have lower total recognition and a higher number of false positives on all three recognition subtests compared to those with either PD or ET. We further expected that (2) individuals with PD would perform similarly to those with ET across recognition scores. Finally, we hypothesized that (3) a greater benefit from recognition cues compared to delayed free recall performance would be demonstrated in the PD and ET samples than in the mild AD sample. Further validation of these new recognition subtests would add to their potential clinical utility. Moreover, enhancing recognition memory assessment offers the potential to improve recommendations for patients with memory concerns by determining the benefit provided by cues for different types of information.

Method

Participants

The present study had a total of 196 participants: 36 participants with mild AD, 55 participants with PD, and 105 participants with ET. The mild AD sample was recruited from senior centers, independent living facilities, and a cognitive disorders clinic to participate in a larger research study. Mild AD was classified based on the Alzheimer’s Disease Neuroimaging Initiative (ADNI) criteria for probable AD, which has been previously described (Duff et al., 2021). Participants with mild AD were excluded if they had a history of significant brain injury (e.g., stroke, brain injury with ≥30 minutes of loss of consciousness), neurological or psychiatric illness, antipsychotic or anticonvulsant medications, substance abuse, score >6 on the Geriatric Depression Scale (15-item version), allergies, motor/sensory impairments, or other concerns that would interfere with completing study procedures. Additionally, medical records were reviewed to ensure that included participants did not have non-AD pathologies (e.g., Parkinson’s disease). The PD and ET participants were recruited from an academic medical center’s Movement Disorders Clinic, where they were clinically diagnosed with PD, ET, or ET Plus (Bhatia et al., 2018) and were considering deep brain stimulation/focused ultrasound (DBS/FUS) for tremor treatment. Notably, patients considering DBS/FUS are a more selective group and may not represent the general population of PD and ET patients. However, understanding the baseline cognitive functioning of this subpopulation is more critical given that these individuals are considering potentially invasive procedures that could result in potentially adverse cognitive effects. Participants with PD or ET were included if they were clinically diagnosed with one of these conditions by a movement disorders specialist, and they were excluded if they had neurological deficits associated with a different movement disorder (e.g., dystonia, drug-induced tremor).

Measures

Repeatable Battery for the Assessment of Neuropsychological Status (RBANS)

The RBANS is a brief cognitive test battery that assesses immediate memory, visuospatial/constructional skills, language, attention, and delayed memory (Randolph et al., 2012). A Total Scale score is derived from the Index scores for these 5 cognitive domains, which are based on the performance across the 12 subtests. Notably, the Delayed Memory Index is comprised of List Recall, List Recognition, Story Recall and Figure Recall subtests. Age-corrected standard scores (M = 100, SD = 15) are obtained for both Index and Total Scale scores. The RBANS normative sample is comprised of 540 cognitively intact individuals aged 20–89 years old. The RBANS has demonstrated adequate internal consistency and test-retest reliabilities, although this is greater for Index compared to subtest scores (Duff et al., 2005; Randolph et al., 2012).

Story and Figure Recognition Subtests.

As described by Duff and colleagues (2021), recognition trials were developed for the story and figure components in the RBANS to be consistent with that for the list component (i.e., List Recognition subtest). Accordingly, both verbal and visual recognition memory are evaluated with the addition of these subtests. Such a comprehensive assessment of recognition memory can better inform whether cueing is a beneficial strategy for everyday life or potential cognitive interventions. These novel recognition subtests were administered immediately following the delayed free recall of their respective material. More specifically, the Story Recognition subtest asked the individual to identify 10 valid story components and discriminate them from 10 intermixed distractor components. Similarly, for the Figure Recognition subtest, individuals were required to identify 10 Figure Recall drawing components from 10 distractor components without the context of the larger figure on a flip booklet of 3” x 5” cards. Each subtest provides scores for Total Correct (or the sum of True Positives and True Negatives), Hits (or True Positives), and False Positive Errors.

Procedures

Written informed consent was provided by each participant, or their legally authorized representatives as appropriate, prior to participation in the study. The procedures and use of clinical data for research purposes were approved by the local institutional review board. All participants completed a battery of cognitive tests, which included the RBANS and the two novel recognition subtests. For the mild AD participants, this evaluation was used to assess cognitive functioning and determine study eligibility. For the PD and ET participants, the RBANS was administered under the supervision a board-certified neuropsychologist as part of an evaluation for potential DBS/FUS treatment for tremor. RBANS Form A was administered to all participants and procedures for administration and scoring of subtests were based on the RBANS manual (Randolph et al., 2012).

Data Analyses

The present analyses were conducted using IBM SPSS Statistics (Version 29). Preliminary analyses compared clinical groups on demographic and cognitive variables. More specifically, group differences in demographic variables were analyzed using one-way analyses of variance (ANOVAs) for continuous variables (i.e., age and education) and Pearson chi-squared tests for categorical variables (i.e., gender). Given that the gender proportions differed between groups and the RBANS normative data does not control for gender, group differences in RBANS Index score performances were each examined using two-way ANOVAs. These ANOVAs included the main effects of diagnosis and gender as well as their interaction in order to control for the influence of gender.

Primary Analyses

A nonparametric analysis of covariance (ANCOVA), or Quade’s test, was run to determine whether the mean performance on each recognition subtest score (i.e., Total Correct, Hits, False Positive Errors for List, Story and Figure Recognition) differed between our three clinical groups. Descriptive statistics revealed that all recognition Total Correct and Hits scores were heavily left skewed (Skewness Range: −1.12 - −2.01), except for Figure Recognition Total Correct scores (Skewness = −.44). A significant right skew was observed for List Recognition (Skewness = 2.17) and Story Recognition (Skewness = 1.29) False Positive scores, while Figure Recognition False Positive scores were not skewed (Skewness = .57). Additionally, most recognition score distributions were more heavily tailed than a normal distribution (Kurtosis Range: 1.19 – 5.73), with the exception of Story Recognition Total Correct scores (Kurtosis = .84). Thus, nonparametric analyses were selected. Further, due to differences in inclusion and exclusion criteria, our samples differed significantly on some demographic variables (e.g., age, gender) that need to be controlled for. For significant Quade’s test results, pairwise comparisons were run to determine specific group differences. The alpha level cutoff was set to .01 for pairwise comparisons to reduce the risk of a Type I error. The percentage of non-overlapping scores between pairs of the three groups was calculated for each recognition score (e.g., number of participants with AD at a given score but no participants with PD at that same score) and used to approximate Cohen’s d (Cohen, 1988). These values, which are estimates of group difference effect sizes for the recognition scores, may provide clinicians and researchers with more information about the relative utility of these recognition scores.

Secondary Analyses

Given the typical discrepancy between recall and recognition that differentiates cortical from subcortical presentations, we conducted a series of analyses comparing free recall to recognition subtest performances for the three diagnostic groups. More specifically, related-samples Wilcoxon signed rank tests were used to compare score differences due to the nonparametric nature of our data. Of the recognition scores, Recognition Hits were determined to be the most comparable recognition score to Recall scores due to the inclusion of only correctly identified information (i.e., neither score is influenced by false positive errors). Given differences in the total number of points possible between subtests (e.g., Story Recall has a maximum of 12 points, but Story Recognition Hits has a maximum of 10 points), the recall and recognition Hits scores were converted to percentages of total possible points for List, Story, and Figure subtests. For example, 9 out of a total possible of 12 points for Story Recall would be 75%, but 9 out of 10 for Story Recognition Hits would be 90%.

In light of the support for the diagnostic utility of recognition discriminability measures (Goldstein et al., 2019; Russo et al., 2017), Quade’s tests were run for a discriminability index in the current sample to determine whether the groups differed in their ability to distinguish between target and distractor items. More specifically, for each recognition subtest (i.e., List Recognition, Story Recognition, and Figure Recognition), the discriminability index was calculated by subtracting the raw False Positive Error score from the raw Hit score. As before, pairwise comparisons with an alpha level cutoff of .01 were run to determine specific group differences for significant Quade’s test results. A discriminability index of 0 reflects performance at chance level while a more positive discriminability index suggests better discrimination of targets from distractors and a more negative discriminability index suggests poorer discrimination of targets from distractors.

Results

Preliminary Analyses

The demographics of our three samples are presented in Table 1. ANOVA results indicated that age differed between the clinical groups (F(2,193) = 14.61, p < .001), with the individuals with mild AD being significantly older than the individuals with ET (p < .001), who were significantly older than those with PD (p < .05). The groups statistically differed with respect to gender proportions (χ²(2) = 6.50, p < .05), with a greater proportion of females in the mild AD group (all p-values < .05). The overall sample was generally highly educated (Years of Education: M = 15.32, SD = 2.49), and this did not differ between groups (p > .05). With regards to RBANS performances, the PD and ET groups were comparable across all RBANS Indexes (all p-values > .05), but the individuals with mild AD had poorer performances compared to the other two groups for the Immediate Memory, Language, Delayed Memory, and Total Scale scores (all p-values < .001), when controlling for the effect of gender and the interaction between diagnosis and gender. Across the sample, females (vs males) had better performances for the Immediate Memory (F(1,190) = 4.12, p = .044) and Language (F(1, 190) = 6.29, p = .013) Indexes. Notably, the interaction between diagnosis and gender was not significant for any RBANS Index score comparison (all p-values > .05).

Table 1.

Descriptive Statistics for Sample Demographics and RBANS Indexes

Demographics	AD (n = 36)	PD (n = 55)	ET (n = 105)	F(df)	p-value	Direction of differences
Age	76.31 (6.39) [66–91]	66.33 (8.44) [40–81]	69.97 (9.34) [39–87]	14.61 (2, 193)	<.001	AD > ET > PD
Education	15.94 (2.27) [12–20]	15.13 (2.42) [12–20]	15.21 (2.59) [8–20]	1.40 (2,193)	.248
Gender	55.6	30.9	34.3	6.50† (2)	.039	AD > PD, ET
RBANS Index Score
Immediate Memory	66.19 (13.91) [40–100]	91.18 (14.93) [57–126]	94.70 (15.57) [57–136]	52.22 (2, 190)	<.001	PD, ET > AD
Visuospatial/Constructional	91.97 (17.19) [60–121]	100.51 (18.05) [62–136]	99.19 (16.94) [64–136]	2.11 (2, 190)	.124
Language	81.86 (14.65) [40–110]	97.24 (10.28) [68–120]	97.13 (10.76) [75–129]	30.07 (2, 190)	<.001	PD, ET > AD
Attention	90.11 (16.55) [49–125]	92.25 (14.73) [64–128]	88.39 (14.15) [49–125]	.88 (2, 189)	.419
Delayed Memory	52.58 (11.62) [40–82]	94.27 (14.04) [56–127]	97.13 (14.98) [52–126]	137.28 (2, 190)	<.001	PD, ET > AD
Total Scale	69.31 (13.55) [40–97]	93.07 (13.17) [62–120]	93.72 (14.22) [63–121]	44.53 (2, 189)	<.001	PD, ET > AD

Note. N = 196. For RBANS Index Scores, the F(df) represent the values for the main effect of diagnostic group when controlling for gender and the interaction between gender and diagnostic group. RBANS = Repeatable Battery for the Assessment of Neuropsychological Status; AD = mild Alzheimer’s disease; PD = Parkinson’s disease; ET = essential tremor; Gender = % female.

^†indicates chi-squared statistic values.

Primary Analyses

When controlling for age and gender, the three groups differed across all of the recognition subtests of the RBANS including the novel recognition tasks (all p-values < .01) as presented in Table 2. Based on pairwise comparisons (α = 0.01), individuals with mild AD had the lowest scores for all Total Correct recognition scores (List, Story, and Figure; all p-values < .01). The ET sample had higher Story Recognition Total Correct scores than the PD sample (p < .01), despite the groups being otherwise comparable. Regarding recognition Hit scores, the mild AD sample had the least number of Hits compared to the other groups (List and Story; all p-values < .01), although Figure Recognition Hits were similar to the PD group (p = .037), while the PD and ET samples were comparable (all p-values > .01). As expected, the mild AD group exhibited the greatest number of False Positive Errors (List, Story, and Figure; all p-values < .01). The PD sample had more Story Recognition False Positive Errors than the ET sample (p < .001), despite the groups being otherwise comparable. Overall, the mild AD group had the poorest performance across recognition scores and the ET and PD samples were comparable, with the exception of PD performing worse on Story Recognition scores (i.e., Total Correct, False Positives).

Table 2.

Descriptive Statistics and Group Differences for RBANS Recognition Scores

Recognition Scores	AD (n = 36)	PD (n = 55)	ET (n = 105)	F(df)	p-value	Direction of differences
List Recognition
Total Correct	14.81 (2.47) [8–19]	18.76 (1.35) [15–20]	18.91 (1.49) [13–20]	37.36 (2,193)	<.001	ET, PD > AD
Hits	7.22 (2.31) [1–10]	9.07 (1.10) [5–10]	9.15 (1.22) [3–10]	11.24 (2, 193)	<.001	ET, PD > AD
False Positive Errors	2.42 (1.81) [0–6]	.31 (.69) [0–3]	.24 (.69) [0–4]	41.99 (2, 193)	<.001	AD > PD, ET
d’	4.81 (2.47) [−2–9]	8.76 (1.35) [5–10]	8.91 (1.49) [3–10]	37.36 (2, 193)	<.001	ET, PD > AD
Story Recognition
Total Correct	12.50 (2.38) [6–17]	17.44 (1.69) [13–20]	17.94 (2.01) [9–20]	45.66 (2, 193)	<.001	ET > PD > AD
Hits	7.19 (1.80) [2–10]	8.95 (1.01) [7–10]	8.95 (1.19) [4–10]	11.64 (2, 193)	<.001	ET, PD > AD
False Positive Errors	4.69 (1.82) [0–9]	1.58 (1.38) [0–7]	1.09 (1.55) [0–9]	44.78 (2, 193)	<.001	AD > PD > ET
d’	2.50 (2.39) [−4–7]	7.36 (1.79) [3–10]	7.87 (2.12) [−1–10]	30.76 (2, 193)	<.001	ET, PD > AD
Figure Recognition
Total Correct	12.69 (1.83) [8–16]	15.38 (1.41) [12–20]	15.65 (1.56) [10–20]	30.76 (2, 193)	<.001	ET, PD > AD
Hits	6.89 (2.16) [1–10]	8.24 (.98) [5–10]	8.39 (.94) [4–10]	7.26 (2, 193)	.001	ET > AD
False Positive Errors	4.19 (2.05) [0–8]	2.86 (.99) [0–6]	2.74 (1.16) [0–6]	8.10 (2, 193)	<.001	AD > PD, ET
d’	2.69 (1.83) [−2–6]	5.38 (1.41) [2–10]	5.65 (1.56) [0–10]	43.06 (2, 193)	<.001	ET > PD > AD

Note. N = 196. d’ = discriminability index (i.e., raw Hit score – raw False Positive Error score); RBANS = Repeatable Battery for the Assessment of Neuropsychological Status; AD = mild Alzheimer’s disease; PD = Parkinson’s disease; ET = essential tremor.

As presented in Table 3, based on the percentage of non-overlapping scores between the ET and PD groups, the estimated effect size was very small for all recognition scores (d range = 0.0–0.1). The estimated effect sizes for differences between the mild AD group and both the ET and PD groups were medium to large (d range = 0.6–0.7) on List Recognition Total Correct and discriminability scores. Large effect sizes were observed for Story Recognition Total Correct and discriminability score differences between the mild AD group and both the ET and PD groups (d Range = 0.9–1.2). Figure Recognition Total Correct and discriminability score differences between the mild AD group and both the ET and PD groups were associated with small to medium effect sizes (d Range = 0.2–0.4). All other between-group score difference effect sizes were very small (i.e., <0.2).

Table 3.

Percent Overlap between Groups and Associated Effect Size for RBANS Recognition Scores

	ET vs. PD		ET vs. AD		PD vs. AD
Recognition Scores	% Non-overlap	d	% Non-overlap	d	% Non-overlap	d
List Recognition
Total Correct	2	0.0–0.1	40	0.6–0.7	41	0.6–0.7
Hits	1	0.0–0.1	2	0.0–0.1	8	0.1–0.2
False Positive Errors	1	0.0–0.1	4	0.0–0.1	11	0.1–0.2
d’	1	0.0–0.1	40	0.6–0.7	41	0.6–0.7
Story Recognition
Total Correct	1	0.0–0.1	60	1.1–1.2	53	0.9–1.0
Hits	4	0.0–0.1	1	0.0–0.1	14	0.1–0.2
False Positive Errors	2	0.0–0.1	8	0.1–0.2	7	0.0–0.1
d’	2	0.0–0.1	59	1.1–1.2	53	0.9–1.0
Figure Recognition
Total Correct	5	0.0–0.1	25	0.3–0.4	20	0.2–0.3
Hits	1	0.0–0.1	4	0.0–0.1	4	0.0–0.1
False Positive Errors	0	0.0	2	0.0–0.1	3	0.0–0.1
d’	5	0.0–0.1	25	0.3–0.4	20	0.2–0.3

Note. N = 196. AD n = 36. PD n = 55. ET n = 105. Cohen’s d was estimated based on conversion of the percentage of non-overlapping scores as an approximation of effect size. d = approximate Cohen’s d value; % Non-overlap = the percentage of non-overlapping scores between the groups. d’ = discriminability index (i.e., raw Hit score – raw False Positive Error score); RBANS = Repeatable Battery for the Assessment of Neuropsychological Status; AD = mild Alzheimer’s disease; PD = Parkinson’s disease; ET = essential tremor.

Secondary Analyses

The results of the related-samples Wilcoxon signed rank tests for List, Story, and Figure recall and recognition Hits are presented in Table 4, Table 5, and Table 6 respectively. Overall, recognition Hits were significantly better than recall scores across groups and subtests (all p-values < .001), with the exception of comparable Figure Recall and Recognition Hits for individuals with PD (p > .05).

Table 4.

Descriptive Statistics and Differences between RBANS List Recall and Recognition Hits Across Groups

Diagnostic Group	List Recall	List Recognition Hits	Subtest Difference	p-value	Direction of differences
AD (n = 36)	.06 (.17) [.00–1.00]	.72 (.23) [.10–1.00]	5.24	<.001	Hits > Recall
PD (n = 55)	.46 (.26) [.00–.90]	.91 (.11) [.50–1.00]	6.47	<.001	Hits > Recall
ET (n = 105)	.45 (.27) [.00–1.00]	.92 (.12) [.30–1.00]	8.79	<.001	Hits > Recall

Note. N = 196. All scores were percentages derived from earned points/total possible points. The Subtest Difference is the z-score value from the Wilcoxon signed rank test. Hits = List Recognition Hits; RBANS = Repeatable Battery for the Assessment of Neuropsychological Status; AD = mild Alzheimer’s disease; PD = Parkinson’s disease; ET = essential tremor.

Table 5.

Descriptive Statistics and Differences between RBANS Story Recall and Recognition Hits Across Groups

Diagnostic Group	Story Recall	Story Recognition Hits	Subtest Difference	p-value	Direction of differences
AD (n = 36)	.13 (.12) [.00–.58]	.72 (.18) [.20–1.00]	5.24	<.001	Hits > Recall
PD (n = 55)	.67 (.19) [.25–1.00]	.90 (.10) [.70–1.00]	6.20	<.001	Hits > Recall
ET (n = 105)	.69 (.22) [.08–1.00]	.90 (.12) [.40–1.00]	8.35	<.001	Hits > Recall

Note. N = 196. All scores were percentages derived from earned points/total possible points. The Subtest Difference is the z-score value from the Wilcoxon signed rank test. Hits = Story Recognition Hits; RBANS = Repeatable Battery for the Assessment of Neuropsychological Status; AD = mild Alzheimer’s disease; PD = Parkinson’s disease; ET = essential tremor.

Table 6.

Descriptive Statistics and Differences between RBANS Figure Recall and Recognition Hits Across Groups

Diagnostic Group	Figure Recall	Figure Recognition Hits	Subtest Difference	p-value	Direction of differences
AD (n = 36)	.07 (.09) [.00–.30]	.69 (.22) [.10–1.00]	5.24	<.001	Hits > Recall
PD (n = 55)	.73 (.23) [.25–.95]	.82 (.10) [.50–1.00]	1.59	.111	No difference
ET (n = 105)	.74 (.19) [.15–1.00]	.84 (.09) [.40–1.00]	4.31	<.001	Hits > Recall

Note. N = 196. All scores were percentages derived from earned points/total possible points. The Subtest Difference is the z-score value from the Wilcoxon signed rank test. Hits = Figure Recognition Hits; RBANS = Repeatable Battery for the Assessment of Neuropsychological Status; AD = mild Alzheimer’s disease; PD = Parkinson’s disease; ET = essential tremor.

As presented in Table 2, the discriminability index differed across the three groups for all of the recognition subtests when controlling for age and gender (all p-values < .01). Subsequent pairwise comparisons (α = 0.01) revealed that the mild AD group had lower discriminability indexes, and thus poorer discrimination between target and distractor items, than either the PD or the ET groups across all recognition subtests (all p-values < .01). The ET and PD groups had comparable discriminability indexes for List and Story Recognition subtests (both p-values > .01), although the PD group had a significantly lower discriminability index on the Figure Recognition subtest (p < .01).

Discussion

To further evaluate the clinical utility of the newly developed RBANS recognition subtests, we compared performances in populations considered classically “cortical” (i.e., AD) and “subcortical” (i.e., PD), along with a clinical comparison (i.e., ET), while controlling for age and gender. Our initial hypotheses were supported in that individuals with mild AD had the lowest performances on all recognition subtests, except for comparable Figure Recognition Hits to the PD sample. Largely consistent with expectations, individuals diagnosed with ET had largely comparable performances to the PD sample across recognition subtests, except for better performance on Story Recognition Total Correct driven by lower False Positive Errors. Further, effect size estimates based on the percentage of non-overlapping scores suggest that ET and PD groups are comparable. In general, of the recognition scores, Total Correct and the discriminability index had comparatively less overlap when comparing the mild AD group to the PD and ET groups. Story Recognition appeared to be the most discriminating subtest in the present sample as the Recognition Total Correct and discriminability index scores demonstrated the lowest percent overlap between the mild AD group and both the PD and ET groups. Overall, these results support the differentiation between cortical and subcortical profiles on RBANS recognition scores, which is consistent with existing literature on recognition performances on other memory tests in similar types of clinical populations (Brandt et al., 1992; Salmon & Bondi, 2009; Salmon & Filoteo, 2007).

Importantly, the observed recognition scores offer potential for clinically meaningful interpretation. Most notably, individuals with mild AD appear to demonstrate around two times as many false positive errors as individuals with either PD or ET. This pattern was apparent regardless of recognition subtest despite considerable percent overlap in scores between all groups. Although the differences in performance between these clinical groups offers insight into different profiles, it is vital to also understand how they relate to intact individuals. In the initial validation study for the novel recognition subtests, Duff et al. (2021) provided preliminary normative means and standard deviations for List, Story, and Figure Recognition scores from a cognitively intact sample. When the scores from the present sample were z-transformed based on the mean and standard deviation from the intact sample (Duff et al., 2021), the recognition impairment of the mild AD sample is clear (see Table 7 for descriptive statistics of all z-transformed recognition scores). More specifically, all recognition scores were far below 1.5 standard deviations from the mean with the worst average performance on List Recognition Total Correct (M = −8.16, SD = 4.12) and best average performance on Figure Recognition False Positive Errors (M = 1.71, SD = 1.47, please note that higher z-scores on this score indicate worse performance). Additionally, the recognition scores generally fell within one standard deviation for the ET and PD samples, except for the average List Recognition Total Correct z--score falling just over 1.5 standard deviations below the mean in the PD sample. Thus, the ET and PD samples were weaker than an intact sample on these subtests, but generally not at a level of clinical impairment. Due to the skew of the scores, preliminary normative percentiles from the cognitively intact sample were provided for the recognition subtest scores (Duff et al., 2021). Based on these percentiles, all average recognition scores from each group were in the <25^th percentile range, except for Figure Recognition Hits in the 25–50^th percentile range in the PD and ET samples. Therefore, the percentile bands appear less sensitive than the means and standard deviations in differentiating the level of recognition impairment within the present clinical samples. Taken together, from a clinical perspective, the observed performances suggest a weakness in recognition for ET and PD and a recognition impairment in mild AD on these across the three recognition subtest scores.

Table 7.

Descriptive Statistics for RBANS Recognition z-Scores Based on Preliminary Intact Sample from Duff and colleagues 2021

Recognition Scores	AD (n = 36)	PD (n = 55)	ET (n = 105)
List Recognition
Total Correct	−8.16 (4.12) [−19.50 – −1.17]	−1.56 (2.24) [−7.83 – .50]	−1.31 (2.48) [−11.17 – .50]
Hits	−4.96 (4.61) [−17.40 – .60]	−1.25 (2.21) [−9.40 – .60]	−1.10 (2.43) [−13.40 – .60]
False Positive Errors	5.79 (4.53) [−.25 – 14.75]	.52 (1.73) [−.25 −7.25]	.35 (1.72) [−.25 – 9.75]
Story Recognition
Total Correct	−4.57 (1.70) [−9.21 – −1.36]	−1.05 (1.20) [−4.21 – .79]	−.68 (1.43) [−7.07 – .79]
Hits	−2.76 (2.25) [−9.25 – .75]	−.57 (1.26) [−3 – .75]	−.56 (1.48) [−6.75 – .75]
False Positive Errors	4.55 (2.02) [−.67 – 9.33]	1.09 (1.51) [−.67 – 7.11]	.54 (1.72) [−.67 – 9.33]
Figure Recognition
Total Correct	−2.71 (1.08) [−5.47 – −.76]	−1.13 (.83) [−3.12 – 1.59]	−.97 (.92) [−4.29 – 1.59]
Hits	−2.01 (2.16) [−7.90 – 1.10]	−.66 (.98) [−3.90 – 1.10]	−.51 (.94) [−4.90 – 1.10]
False Positive Errors	1.71 (1.47) [−1.29 – 4.43]	.75 (.71) [−1.29 – 3]	.67 (.83) [−1.29 – 3]

Note. N = 196. Scores are z-transformed raw scores for each recognition subtest based on the preliminary means and standard deviations from the intact sample in Duff et al. 2021 (see Table 1 for intact sample means and standard deviations for all recognition subtests). The respective mean from the intact sample was subtracted from each participant’s raw score in the current study and then divided by the intact sample standard deviation. A lower z-score indicates poorer performance compared to the intact sample for all recognition scores, except for False Positive Errors where a higher z-score indicates poorer performance (i.e., more errors). RBANS = Repeatable Battery for the Assessment of Neuropsychological Status; AD = mild Alzheimer’s disease; PD = Parkinson’s disease; ET = essential tremor.

Our secondary goal was to determine whether differences between retrieval and recognition were present within the clinical groups. On the one hand, the mild AD sample exhibited the highest amount of false positive errors, which supports the notion that the recognition impairment in AD is the result of a failure to discriminate between targets and distractors (Brandt et al., 1992; Clark et al., 2012; Hildebrandt et al., 2009). However, contrary to expectations, we found that all groups had better recognition based on percent of target hits compared to percent freely recalled, with the only exception being comparable performances for Figure Recognition in the PD sample. The better recognition performance in comparison to free recall observed for the mild AD sample could be interpreted as this group benefitting from recognition, which would be contrary to what is expected with a “cortical” presentation (Arango-Lasprilla et al., 2006; Salmon & Filoteo, 2007). Indeed, the lack of complete separation between groups could be indicative of mixed presentations (i.e., both “cortical” and “subcortical” pathologies; Arango-Lasprilla et al., 2006; Bonelli & Cummings, 2008). However, it is worth noting that the percent recalled in this group was near 0 across recall subtests (Mean Percent Recalled Range: .06 - .13) and their recognition Hit scores were significantly lower than the other two groups. Therefore, the “benefit” observed for recognition compared to recall in the mild AD sample is more reflective of greater within group change despite poorer performance overall. For example, in the mild AD sample the difference between the average percent List Recall (M = .06) and percent List Recognition Hits (M = .72) is .66, whereas this difference is .45 for the PD sample and .47 for the ET sample. Thus, although the mild AD sample had a greater difference, their percent accuracy for List Recognition Hits was around 20% lower than the PD (M = .91) and ET (M = .92) samples. Moreover, the mild AD group demonstrated the poorest discrimination between target and distractor items across all recognition subtests. Taken together, the mild AD sample did appear to benefit from recognition cues, but not to the extent that their scores were suggestive of intact recognition.

Even though the mild AD group was included as a classic “cortical” condition and the PD group was included as a classic “subcortical” condition, the inclusion of the ET group deserves some additional explanation. The ET group was intended to act as an additional comparison that is traditionally considered “subcortical,” but has been associated with heterogeneous cognitive presentations (Ratajska et al., 2022). The ET sample in the present study exhibited Index and recognition subtest scores were largely comparable to the PD sample. This similarity is unsurprising as ET is increasingly associated with cognitive decline (Louis, 2016). Cognitively, individuals with ET typically exhibit mild impairments on measures of executive functioning, attention/working memory, and memory (Bermejo-Pareja & Puertas-Martín, 2012; Janicki et al., 2013; Shanker, 2019). Notably, individuals with ET have been shown to exhibit weaknesses on Immediate Memory, Attention, and Language Indexes of the RBANS (Vehar et al., 2023). Within the memory domain, recognition of word lists in ET samples tends to be significantly lower than normative samples or controls (Puertas-Martín et al., 2016; Şahin et al., 2006; Sinoff & Badarny, 2014; Tröster et al., 2002; but see Kim et al., 2009); however, it also appears comparable to individuals with PD (Lacritz et al., 2002; Lombardi et al., 2001; Puertas-Martín et al., 2016) and not at a rate greater than expected in a normal distribution (Tröster et al., 2002) suggesting a subcortical profile. Performance for visual memory recognition measures is mixed (Kim et al., 2009; Tröster et al., 2002) and to our knowledge there is no explicit investigation of verbal story memory recognition. The observed similarities between the ET and PD samples in the present study are consistent with a more subcortical profile of ET across verbal and visual recognition measures and further argue against ET as a cognitively intact condition.

Several limitations of the present study should be considered. First, the smaller sample sizes for the mild AD and PD groups increases the chance that some effects may be over-estimated and/or that some analyses may be underpowered (Button et al., 2013). Second, the generalizability of the present findings to the larger population of these neurological groups may be limited by certain characteristics of the included samples. More specifically, there was a high level of educational attainment across the samples and the ET and PD samples were seeking procedural treatment for their tremors (i.e., DBS or FUS). Additionally, data regarding race and ethnicity of participants was not obtained for all participants. Future studies should take efforts to include larger samples with greater diversity of demographic characteristics and treatment considerations. Third, the different recruitment methods for the cortical and subcortical samples warrants consideration. To enhance generalizability to populations referred for outpatient neuropsychological assessment, future research should include representative samples from that setting. It is also worth noting that given the inclusion criteria for the PD sample, it cannot be excluded that some of the individuals may have co-occurring AD pathology (Tropea et al., 2023). Indeed, such neuropathological overlap is common and suggests a continuum between “cortical” and “subcortical” pathology (Bonelli & Cummings, 2008) that may explain the lack of complete separation between groups in the present study. Fourth, it cannot be ruled out that the ET sample results were influenced by within-group heterogeneity given the inclusion of individuals diagnosed with ET Plus, which can include a neurological “soft sign” of memory impairment (Bhatia et al., 2018). Although ET Plus is not a universally accepted diagnosis at this time (Louis, 2020), determining differences in cognitive profiles between ET and ET Plus remains an important avenue for future research. Fifth, suboptimal effort in our samples is unlikely, but cannot be ruled out as a formal assessment of performance validity was not conducted with the current sample. Although approximately 7% of our participants (0 PD, 2 ET, 11 AD) exceeded the cutoff of 3 for the embedded RBANS Effort Index (EI; Silverberg et al., 2007), these failures were exclusively driven by poorer performance on the List Recognition subtest (vs. the Digit Span subtest). There were significant correlations between EI scores and Total Scale scores for each group (ET: r = −0.41, p < .001; PD: r = −0.43, p = .001; AD: r = −0.53, p = .001). Thus, the observed low EI scores are interpreted as the result of true cognitive impairment in the participants and not poor effort. Sixth, the results presented should be considered in light of the psychometrics of the newly developed Story Recognition and Figure Recognition subtests. In particular, the correlations observed across the recall and recognition subtests were shown to be moderately large (rho = .57–.92), and Story Recognition exhibited greater internal consistency (Cronbach’s alpha = 0.78; split-half coefficient = 0.76) than Figure Recognition (Cronbach’s alpha = 0.58; split-half coefficient = 0.61; Duff et al., 2021). Of note, the Figure Recognition subtest may be limited to one aspect of figure recognition (i.e., elements) as it does not assess recognition of the appropriate placement of drawing elements. Relatedly, the present study did not directly examine the incremental validity of including the newly developed recognition subtests (i.e., Story Recognition, Figure Recognition). It will be important for future studies to establish whether adding these additional subtests to the RBANS battery provides clinically meaningful and unique information over and above what the current subtests provide with a variety of clinical populations.

Despite the noted limitations, our study is the first to demonstrate clinically useful differences on the novel RBANS recognition subtests. More specifically, the mild AD sample performed more poorly across recognition scores (i.e., total, hits, false positive errors) than both PD and ET, consistent with the traditional differentiation between cortical and subcortical memory profiles. Additionally, at a within-group level, all clinical samples exhibited better performance on recognition than free recall of the respective material. Thus, our findings encourage the inclusion of these recognition subtests in future clinical practice and research endeavors.