Abstract
The Booklet Category Test (BCT) is a neuropsychological test of cognitive dysfunction that provides only one overall error score indicative of global impairment. It does not, however, delineate specific domains that might be impaired. The aim of this study is to concurrently validate 13 new BCT subscales using legacy instruments in patients with nonpenetrating traumatic brain injury (TBI). Eighty-nine patients with mild, moderate, and severe TBI completed a battery of neuropsychology tests. Partial correlations controlling for age were performed and there were significant correlations between the a priori selected scores from legacy measures of major cognitive domains and both BCT total errors and subscale scores. Additional analysis showed that several subscales were able to differentiate between performance levels on the legacy measures. Overall, our results showed that the subscales measured cognitive skills beyond global impairment, supporting the use of the BCT subscales in a population with TBI.
Keywords
The Booklet Category Test (BCT) (DeFilippis & McCampbell, 1997) is a well-recognized tool in the assessment of brain dysfunction (Goldstein & Shelly, 1972; Reitan & Wolfson, 1993, 1995). While it was originally thought to be sensitive to impairments resulting from frontal lobe damage, subsequent studies have demonstrated that the test is actually an indicator of global impairment (Anderson, Bigler, & Blatter, 1995). The BCT’s generation of a single error score has been the subject of criticism since Simmel, Counts, and Pressey (1957) proposed that the original Halstead (1947) Category Test (CT) should be viewed as a multidimensional instrument requiring various cognitive skills. Moreover, McCune, Peck, Mitchell, and Ellwood (1992) found that the CT was more related to sequential problem solving with more intact individuals, whereas it was more strongly related to visual and memory abilities in individuals with more severe brain injury. Along with the review by Choca, Laatsch, Wetzel, and Agresti (1997), these results suggest that the BCT has the capability to provide insight into specific underlying deficits that contribute to overall brain dysfunction.
In the past, researchers have tried to expand the utility of the BCT by attempting to identify different cognitive domains that are assessed by the BCT through the use of factor analysis scores (Allen, Goldstein, & Mariano, 1999; Donders & Kirsch, 1991; Johnstone, Holland, & Hewett, 1997). Additional subscores have been shown to not only be efficacious in discerning between patients with and without brain damage but also between different traumatic brain injury (TBI) severities (Allen, Caron, Duke, & Goldstein, 2007; Donders, 2001).
Attempts to develop limited subscales for the BCT have yielded a variety of methods to measure different cognitive domains thought to be captured by the BCT (Minassian, Perry, Carlson, Pelham, & DeFilippis, 2003; Schutz & Gorman, 1995; Webster & Lopez, 2006). Recently, Jorge Costa, at the University of Aveiro, developed a set of 13 subscales using both existing literature and new approaches to fully represent all major cognitive domains relevant to the BCT (personal communication). The new scales that Costa developed were initially based on those used by Minassian et al. (2003) in their research. These scales were refined and new ones were developed by DeFilippis and colleagues during several applications of the scales in a number of clinical populations. Scales that did not appear to relate to measures of similar constructs were modified, and a new Category scale was developed. Costa and colleagues further modified the final scales to have them correspond more closely to their intended constructs. This was a bootstrapping technique (Cronbach & Meehl, 1955) in which various versions of the scales were correlated with known measures of constructs, and the modifications were implemented to improve correlations.
The objective of the present study is to perform a concurrent validity study of the 13 subscales developed by Costa by comparing them with the commonly used instruments. It was hypothesized that the scores for BCT subscales would have significant correlations with the appropriate corresponding measures.
Method
Participants
In accordance with the National Institutes of Health Institutional Review Board, participants in this study were enrolled in a prospective cohort study of nonpenetrating TBI at the National Institutes of Health. Patients were either in a longitudinal arm of the study, in which they were enrolled within 1 year of injury and then followed through up to 9 time points, or a cross-sectional substudy consisting of one visit.
Participants were eligible for the study if they were 18 years of age or older, diagnosed with a nonpenetrating mild, moderate, or severe TBI using the Department of Veterans Affairs/Department of Defense (2009) TBI severity rating criteria. Participants were excluded if they had a contraindication to magnetic resonance imaging scanning, an inability to read or speak English at a level necessary to complete the clinical phenotyping assessments, or medical or psychological instability causing the subject to not reasonably be able to complete the study requirements. Proxy consents were allowed for those who could not consent on their own.
Measures
The legacy instruments compared to the BCT subscales were selected based on their similarity to the proposed domain measured by the BCT subtest in question (Table 1). The Test of Premorbid Functioning (Pearson, 2009) was used to estimate the IQ of individuals, and thus attain some indication of ability, prior to head injury.
Assessments Used to Validate BCT Subscales.
Booklet Category Test
The BCT consists of 208 items, which are divided into seven subtests. The nonverbal stimuli are presented to the patient, with each item being on a separate page, and the patient indicates a number between 1 and 4 that they think the geometric figure or design suggests. After each response, the patient gets feedback on whether they were correct so that they can adjust their strategy appropriately. The patient is also informed when the subtest ends and a new one begins, and is reminded that the principle may be the same as in the last test or that it may change. Each new stimuli item the patient sees is novel, with the exception of Subtest 7, which has items from previous subtests, and the stimulus type may change within a subtest. For example, the stimuli in a subtest may change from triangles, to lines, to diamonds, but the patient should continue to use the same principle to respond to each item.
The 13 New BCT Subscales
Category (CAT-2)
The Category subscale is a measurement of how quickly an individual is able to identify and learn a category. This score takes into account the number of trials it takes for an individual to successfully identify the concept for three consecutive items. Once the individual does this, CAT-2 is established. However, if the individual then makes an error and is not able to successfully pick the category back up within three items, it is assumed that the test taker only coincidentally provided the correct responses. CAT-2 resets and continues to assess the trials it takes until the individual can identify the concept, again, by giving the correct response to three consecutive items. An individual whose learning ability is impaired would take more trials before determining the concept, and correctly applying it throughout the rest of the subtest.
Category (CAT-2A)
Like CAT-2, this subscale also measures learning by assessing the number of trials an individual requires before being able to successfully identify and apply the concept. However, CAT-2A will not reset for individuals who correctly identify 10 consecutive items before making an error.
Set Loss (SL-A, SL-C, and SL-T)
Set loss errors are errors made after the individual has established CAT-2. Mistakes made after an individual presumably knows how to respond to each item are not likely to be random errors, nor are they reflective of difficulty with determining the correct concept for the items. SL-A provides indication of errors that were made due to a sudden lapse of attention or impulsivity. If an individual successfully responds to consecutive items, all with the same concept and stimulus structure, it is thought that an ensuing error is attentional rather than conceptual. SL-C is a measurement of concept formation, abstraction, mental flexibility, and susceptibility to confusion. A score for this subscale is obtained by counting the errors made when the core stimulus changes between subgroups, within a subtest. SL-T is the sum of SL-A and SL-C scores and is a measurement of overall loss of set.
Spatial Positional and Proportional Reasoning (SPR and PR)
The Spatial Positional Reasoning score is derived from the errors made on Subtests III, IV, and VII. Proportional Reasoning is based on errors made on Subtests V and VI. The development of these subscales was based on previous factor analysis studies that indicated spatial and proportional reasoning abilities were measured by the BCT and the respective subtests loaded onto the factors (Allen et al., 1999; Johnstone et al., 1997).
Memory (M)
Principles from Subtests 1 to 6 are presented on Subtest 7, making performance on Subtest 7 an indicator of memory. The M score is derived using Minassian et al.’s (2003) scoring system. The score is determined when an item on Subtest 7 was both correct and met the criterion (i.e., the subtest the item is from had five consecutive correct answers), then that is divided by the number of items, whether correct or incorrect, on Subtest 7 that met the five item correct criterion (Minassian et al., 2003).
Perseveration (PSV)
The examinee receives feedback after each item on the BCT, so an examinee who continues to provide the same response or apply the same principle is perseverating. The way in which an individual perseverates, however, can vary due to the complex and diverse stimuli presented in the BCT. If the same response is continuously given (1-1-1-1), the perseveration is described as true perseveration (PSV-T) and each response counts toward the PSV-T score, regardless of whether it was correct or incorrect. An examinee may also perseverate between subtests (PSV-B). Perseveration of this type occurs when the concept from the previous subtest is applied to the next subtest despite attaining feedback that the principle is incorrect. If a patient continues to apply the same concept to every item in a subtest, without attempting to adjust after receiving feedback, they are perseverating within a test (PSV-W). This type of perseveration is more concrete in nature; for example, a patient persists on counting only solid elements of each design after receiving feedback that the approach is incorrect. When a patient perseverates within a subtest, they may count the number of similar aspects of each stimulus (PSV-CS) or they might count the parts of the stimulus that are different (PSV-CD).
Procedures
Neuropsychological testing was conducted in time blocks of 2.5 to 3.5 hours for a total of approximately 6 hours. Testing was conducted by trained psychometric technicians and supervised by a neuropsychologist. For the purposes of this study, the longitudinal subjects’ first BCT evaluation was used.
Statistical Analysis
Partial two-tailed correlations controlling for age were performed for each BCT subscale score with its appropriate corresponding test. The corresponding tests were chosen on the basis of generally accepted aspects of cognition that the tests measure (Table 1). Because of the exploratory nature of this study, correlations for which p < .05 were considered significant, and no additional adjustments for multiple comparisons were made.
CAT-2 and CAT-2A were correlated with the CVLT-II List A Free Recall Trials 1-5 Total T(1-5 Total) (Delis, Kramer, Kaplan, & Ober, 1987), WCST Categories Completed T (Categories), and WCST Trials to Complete 1st Category percentile rank (Trials) (Heaton, Chelune, Tallen, Kay, & Curtiss, 1993).
SL-A was correlated with TMT-A T (Reitan, 1979), WAIS-IV Digit Span scaled (Wechsler, 2008), SSPT T, SSRT T, and WCST Failure to Maintain Set percentile rank scores.
SL-C was correlated with TMT-B T (Reitan, 1979), WCST Failure to Maintain Set percentile rank, WAIS-IV Similarities and Matrix Reasoning scaled scores.
SL-T was correlated with WCST Total Error T and Failure to Maintain Set percentile rank scores.
The SPR and PR subscales were compared with the RCFT percentile rank (Meyers & Meyers, 1995), as well as the WAIS-IV Matrix Reasoning and Block Design scaled scores.
The memory subscale, M, was correlated with the CVLT-II Short Delay Free Recall z score and WMS-IV Visual Reproduction scaled scores (Wechsler, 2009).
Each of the perseveration subscales, PSV-T, PSV-CS, PSV-CD, PSV-B, and PSV-W were correlated with the WCST perseverative responses and perseverative errors T scores.
We also examined group differences in the subscale scores between people who were impaired on the legacy measures and those who were not impaired using a one-way ANOVA controlling for age. Individuals classified as impaired in each of the cognitive domains were 1.5 standard deviations below the average on more than half of the tests in that domain. We covaried for age because, while each of the legacy measures are age corrected, the new subscale scores are not.
Results
Eighty-nine adults (56 males, 33 females) were enrolled in this study, and their demographic characteristics can be found in Table 2. Participants had an average age of 44.0 (±16.2) years, 15.3 (±2.6) years of education, and an estimated premorbid IQ (Test of Premorbid Functioning) of 107.3 (±14.3). Average time since injury was 14.2 months.
Demographic Information.
Means and standard deviations for subjects’ performance on the neuropsychological assessments are listed in Table 3. The BCT subscales yielded a number of significant correlations with corresponding assessments, which can be found in Table 4. The overall BCT errors score was also correlated with each of the assessments (Table 5). Additional correlations significant at p < .05 for non–a priori neuropsychological assessments are presented in Table 6.
Means and Standard Deviations for Assessments.
Note. BCT = Booklet Category Test; CVLT = California Verbal Learning Test; WCST = Wisconsin Card Sorting Test; TMT = Trail Making Test; WAIS = Wechsler Adult Intelligence Scale; SSPT = Speech Sounds Perception Test; RCFT = Rey Complex Figure Test; WMS VR = Wechsler Memory Scale Visual Reproduction; SSRT = Seashore Rhythm Test; CAT-2 and CAT-2A = Category; SL-A = Attentional Set Loss; SL-C = Conceptual Set Loss; SL-T = Total Set Loss; SPR = Spatial Positional Reasoning; PR = Proportional Reasoning; M = Memory; PSV-T = True Perseveration; PSV-CS = perseveration using similar items; PSV-B = perseveration between subtests; PSV-CD = perseveration using different items; PSV-W = within subtest perseveration.
Correlations for BCT Subscale Scores and Analogous Neuropsychological Tests.
Note. BCT = Booklet Category Test; CAT-2 and CAT-2A = Category; CVLT = California Verbal Learning Test; WCST = Wisconsin Card Sorting Test; SL-A = Attentional Set Loss; TMT = Trail Making Test; WAIS = Wechsler Adult Intelligence Scale; SSPT = Speech Sounds Perception Test; SSRT = Seashore Rhythm Test; SL-C = Conceptual Set Loss; SL-T = Total Set Loss; SPR = Spatial Positional Reasoning; RCFT = Rey Complex Figure Test; PR = Proportional Reasoning; M = Memory; WMS VR = Wechsler Memory Scale Visual Reproduction; PSV-T = True Perseveration; PSV-CS = perseveration using similar items; PSV-B = perseveration between subtests; PSV-CD = perseveration using different items; PSV-W = within subtest perseveration. Lower scores indicate better performance for CAT-2, CAT-2A, SL-A, SL-C, SL-T, SPR, PR, and the PSV scales. Higher scores indicate better performance for the M scale.
BCT Errors Score Correlations.
Note. BCT = Booklet Category Test; CVLT = California Verbal Learning Test; RCFT = Rey Complex Figure Test; SSPT = Speech Sounds Perception Test; SSRT = Seashore Rhythm Test; TMT = Trail Making Test; WAIS = Wechsler Adult Intelligence Scale; WCST = Wisconsin Card Sorting Test; WMS = Wechsler Memory Scale.
Non a priori legacy measures significantly correlated with BCT subscales at p<.01.
Note. CAT2 & CAT2A = Category; SL-A= Attentional Set Loss; SL-C= Conceptual Set Loss; SL-T = Total Set Loss; SPR = Spatial Positional Reasoning; PR = Proportional Reasoning; M = Memory; PSV-T = True Perseveration; PSV-CS = perseveration using similar items; PSV-CD = perseveration using different items; PSV-B = perseveration between subtests; PSV-W = within subtest perseveration; CVLT = California Verbal Learning Test; RCFT = Rey Complex Figure Test; SSRT = Seashore Rhythm Test; WAIS = Wechsler Adult Intelligence Test; WAIS BD = Block Design; WAIS DS = Digit Span; WAIS MR = Matrix Reasoning; WAIS WCST = Wisconsin Card Sorting Test; WMS VR = Wechsler Memory Scale Visual Reproduction; SSPT = Speech Sounds Perception Test.
Cohen (1992) suggested that a correlation of .1 indicated a small effect size, .3 was moderate, and .5 reflected a large effect size. CAT-2 and CAT-2A scores (lower scores indicate better performance) were found to have moderate negative correlations with CVLT-II 1-5 and WCST Trials to 1st. CAT-2A also had a moderate significant correlation with WCST Categories.
The SL-A scores yielded moderate negative correlations with Trails A and Digit Span; however, no other significant correlations were found with other attention measures. The SL-C and SL-T scores were not significantly correlated with any of the measures.
The SPR and PR scores were significantly negatively correlated with WAIS-IV Matrix Reasoning, WAIS-IV Block Design, and RCFT. Again, the scores for these subscales are derived using error numbers, so a high score signifies poor performance.
Moderate positive correlations for the M subscale were found with WMS-IV Visual Reproduction, but not with the CVLT-II Short Delay Free Recall. For the M subscale, a higher score indicates better performance.
Finally, only the PSV-T and PSV-CD had significant correlations with legacy measures. Lower PSV scores indicate better performance. Both PSV-T and PSV-CD had negative moderate correlations with WCST Perseverative Errors, while only PSV-CD had a moderate negative correlation with WCST Perseverative Responses. None of the other Perseveration scales correlated with either of the perseverative measures.
To determine the subscale scores’ ability to discriminate between individuals who were impaired in a cognitive domain and those who were not impaired, we conducted one-way ANOVAs and covaried for age since the subtests use raw scores, which are not age corrected. Our analysis was significant for CAT-2, F(1, 87) = 14.68, p = .000; CAT-2A, F(1, 87) = 14.77, p = .000; SPR, F(1, 89) = 4.64, p = .034; PR, F(1, 89) = 8.74, p = .004; PSV-T, F(1, 66) = 11.55, p = .001; PSV-CD, F(1, 66) = 5.94, p = .018; and PSV-W, F(1, 66) = 8.91, p = .004.
Discussion
The BCT is widely accepted as a sensitive measure for a variety of neurological disorders. The test owes its wide applicability to the complex set of cognitive functions required to complete the task. However, the single error score generated by the BCT has been criticized for its failure to identify the specific cognitive functions required by the test (Choca et al., 1997). Our results suggest that there may be value in some of the BCT subscales developed by Costa. Their potential to expand the utility of the BCT and advance the understanding of the variety of cognitive skills are supported by the results of this study, which demonstrated significant concurrent validity for many of the proposed subscales. In particular, CAT-2, CAT-2A, SPR, and PR were shown to reflect performances on tests of concept formation, learning, and construction. PSV-T and PSV-CD may also be useful in gaining insight to perseverative tendencies in the BCT.
An integral component of the BCT is one’s ability to deduce the rule by which to categorize each item. While the total number of errors could presumably be related to an individual’s understanding of the categories, they can also be affected by a number of factors, making it difficult to detect the underlying abilities contributing to any variability for those who perform poorly.
The Category subscales (CAT-2 and CAT-2A) scores were associated with more impaired learning on the CVLT-II 1-5 and more trials before grasping a category on the WCST, suggesting that CAT-2 and CAT-2A may be assessing concept formation abilities and that one’s ability to identify a concept and learn the categories in fewer trials is related to better learning overall.
Even though an individual is able to determine a concept and effectively apply it in the task, the patient may still make an error due to problems with attention, abstraction ability, confusion, or mental flexibility. Errors of these types are known as set loss errors (SL). Once the concept has been learned, errors committed due to impaired attention are measured by SL-A. SL-A significantly correlated with Trails A and with WAIS Digit Span, indicating the score may possess some ability to measure attention. The absence of correlations with the other attention measures, however, suggests that before this subscale is used as an indicator of possible attention deficits, it must be further refined. SL-C, a measurement of set loss due to abstraction, mental flexibility, and confusion, was not significantly correlated with any of the legacy measures. While both SL-A and SL-C involve errors made after a concept has been learned, errors due to attention problems are different than errors due to weak concept formation or flexibility. The SL-T subscale score is derived by combining the SL-A and SL-C scores, which are presumably measurements of completely different cognitive problems. This may be a contributory factor to the lack of significant findings between SL-T and the legacy measures. Another possibility pertains to the method by which the set loss scores are calculated. The SL subscales are only calculated once the test taker learns a concept and the CAT scores are established. The scores reflect the items following the point at which the patient demonstrated an understanding of the concept. Consequently, extremely high error scores will affect the SL subscales. Patients who identified the concept almost immediately had low CAT scores and, because they established CAT early in the subtests, had more items with which SL could be calculated. A low CAT score, then, does not necessarily yield a lower or higher SL score, but rather provides more opportunity for deficits to be measured. Patients who took the majority of the subtest to grasp the concept and establish the CAT score (thus, resulting in a high CAT score) had fewer items with which the SL scores could be calculated. Consequently, individuals with similar attention deficits may have different SL-A scores, not because of the scale’s inability to measure attentional mistakes but because one may have had more difficultly with identifying the subtest’s rule. SL cannot, however, be calculated for patients who never identified the concept and established the CAT score. Accordingly, a correction that accounts for the CAT score may provide some benefit to the efficacy of the SL scores.
The stimuli on the BCT are entirely visual, fairly complex, and frequently change, making the assessment of construction and visual reasoning ability especially pertinent. In fact, the SPR and PR subscales had the highest correlations with tests of spatial and proportional reasoning, demonstrating that construction and visuospatial reasoning deficits are manifested in BCT performance.
The M score, developed by Minassian et al. (2003), had significant correlations with the WMS VR1 and CVLT-II. The current subscale calculation does not seem to be able to adequately represent memory impairment, as the M subscale was not able to discriminate between individuals with memory impairments in legacy measures and those who were not. However, there were significant correlations between M and some of the legacy measures, suggesting that it may be beneficial for future studies to improve the memory scale. Additionally, because memory is often intact in TBI patients a year from injury, it may be of use to validate the M scale in a population in which memory problems are not comorbid with other cognitive deficits (i.e., processing speed or attention).
There were some significant correlations for the Perseveration subscales, but only for PSV-CD and PSV-T and WCST perseverative errors, and PSV-CD with WCST perseverative responses. A number of studies have shown that the WCST and the BCT are not directly comparable and that different cognitive processes may be involved in each. While the BCT is more of an indicator of global impairment, Lombardi et al. (1999) found that dorsolateral frontal-subcortical circuit dysfunction was associated with the WCST. Donders and Kirsch (1991) found the BCT and WCST to only be slightly related, with each largely measuring different neuropsychological functions. A study by Perrine (1993) showed that only 30% of common variance was shared between the BCT and the WCST. Furthermore, the types of concept formation measured by the two were different, with the BCT involving rule learning and the WCST involving feature identification. In line with these studies, the present study failed to yield consistent significant findings between the BCT and the WCST perseveration scores, which supports the contention that the concept of perseveration is measured differently in each test. Donders and Kirsch (1991) raised a number of points that illustrate the differences between the WCST and BCT. First, the BCT examiner instructs the patient when it is possible that the concept might change. Additionally, in the WCST the previous four sorts, and thus, elements of the stimuli, are visible to the patient. The BCT not only presents a single stimulus pattern to the examinee at a time, but the stimuli is much more complex and changes throughout the test. This could have an effect on the way perseveration presents in the two tests, providing a possible explanation for the limited correlations observed for the Perseveration subscales. Future studies may seek to validate the scales in a way that accounts for the complexity of the BCT.
The development of these subscales for the BCT has largely been in response to criticism that the single error score received. The overall error score has been shown to be a reliable and valid measure of an individual’s level of neuropsychological impairment; however, the BCT appears to have the potential to provide more specific information. Our results show that the overall BCT error score (T) correlates with almost all our selected measures. While not as extensive as the overall errors score, the subscale scores also had significant correlations with these measures, suggesting that the test is capable of differentially assessing various cognitive abilities. This supports the previous literature indicating that these neuropsychological processes are, in fact, components that contribute to the overall error score and that information about each can be extracted through these subscales. There were significant correlations between the subscales and tests that were not selected a priori; however, these tests generally had some components related to the construct of the BCT subscale.
In addition to exploring the ability of these scales in measuring specific cognitive skills, our study sought to determine if these new scores could discriminate between people who were classified as impaired in specific cognitive domains, based on performance on existing legacy measures. Establishing meaningful subscales for the BCT has value that goes beyond simply the expansion of the test. The results of this study indicate that the BCT has the capacity to provide information about a number of cognitive abilities that are normally assessed using multiple tests. While the cognitive skills assessed by the BCT may be complex and intertwined, measurement of the domains by the subscales show significant group differences between TBI patients who are impaired and unimpaired in abstraction and concept formation, spatial positional reasoning, proportional reasoning, and perseveration. These findings further suggest that these new subscales could be useful in helping inform clinicians about the cognitive domains on which to focus in their evaluations.
This is the first application of these subscales of the BCT in a TBI sample and there are several limitations to the study. First, there was limited range in some of the scores for both the subtests and the legacy measures, which may have affected the correlations. Second, the study sample was not representative of the general U.S. population in that it was not ethnically diverse and had a higher level of education (15 years). This higher level of education may also indicate a high level of intelligence, which has been found to be related to overall level of performance on the BCT. Golden, Kushner, Lee, and McMorrow (1998) found that correcting for IQ yielded a weaker relationship between the BCT and the WCST, which may have been a factor in this study. Additionally, due to the exploratory nature of this study, our alpha level remained at .05.With the exception of set loss, however, we found many of the BCT subscales still stood with legacy measures when an alpha level of .01 is considered.
In order to make the subscales clinically applicable, additional studies need to norm the scores and establish cutoffs for impairment for each of these subscales. Research examining the ability of the subscales to differentiate neurologically impaired patients from healthy individuals is also necessary. Although the correlations between subscale scores and corresponding test scores were generally low, the subscales clearly possess content validity. The positive findings support the continued study of the subscales and their applicability in clinical populations.
Footnotes
Acknowledgements
We would like to thank Sarah Levy and Katherine Lopez for their help in collecting data and Jorge Costa for providing assistance with the BCT subscale scores.
Declaration of Conflicting Interests
The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: Nick DeFilippis is an author and receives royalties for the Booklet Category Test. Shannon McNally, John Dsurney, Jessica McGovern, and Leighton Chan do not have conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Department of Defense (Center for Neuroscience and Regenerative Medicine) and the National Institutes of Health Intramural Research Program.