Elaboration of Neuropsychological Evaluation of Children: Structural Analysis of Test Results

Background Modern neuropsychology is discussing the possibility of combining qualitative and quantitative approaches in the evaluation of cognitive functions. In Russia a battery of tests called “Methods of neuropsychological assessment for children 6–9 years old” (Akhutina et al., 2016) has been proposed; it is based on the Lurian approach to diagnosis and combines qualitative and quantitative approaches to testing. The present paper describes the development of this combined qualitative and quantitative assessment of various groups of cognitive functions in preschool and primary school children. Structural modeling enables us to analyze a possible combination of integral indices of functions that includes the results of both a face-to-face neuropsychological assessment and computerized testing. Objective To develop a combined qualitative and quantitative neuropsychological assessment of children, in order to 1) check the structural reliability of integral indicators of various cognitive functions; and 2) confirm the correctness of combining the results of face-to-face and computerized tests. Design A sample of 299 children between the ages of 6 and 9 years old (111 preschoolers, 82 first graders, and 106 second graders) underwent a Lurian face-to-face neuropsychological examination adapted for 6-to-9 year-old children, and five tests from the Computerized Neuropsychological Assessment for 6–9 Year-old Children. The five were the “Dots” test, the Schulte Tables, the Cancellation test, the Corsi Tapping Block test, and the Understanding of Similar Sounding Words test. In each of the tests (face-to-face and computerized), key parameters were identified to evaluate various cognitive functions. Results A confirmatory factor analysis verified the composition of the neuropsychological indices that were based on the results of the face-to-face neuropsychological assessment. At the same time, when the computer test data were added to the model, the fit indices of the model considerably improved. Conclusion The confirmatory factor analysis confirmed the validity of the identification of eight neuropsychological indices that indicate the component processes underlying complex cognitive functions in children: 1) programming and control of voluntary actions (executive functions); 2) serial organization of movements and speech; 3) the processing of kinesthetic information; 4) the processing of auditory information; 5) the processing of visual information; 6) the processing of visual-spatial information; 7) hyperactivity/impulsivity; and 8) fatigue/slowness.


Introduction
Neuropsychological methods that examine and assess the state of cognitive functions in children are under development all over the world. Earlier stages of the development of such methods were characterized by two di erent approaches: a psychometric quantitative one, and a qualitative one based on the theory of functional systems. e rst approach is based on expert-independent objective testing and quanti cation of results. e second is devoted to a detailed analysis of the quality of the subjects' test performance. However, the present stage of child neuropsychology research is characterized by the convergence of these approaches for determining a diagnosis (Akhutina, Ignat' eva, Maksimenko, Polonskaya, Pylaeva, & Yablokova, 1996;Akhutina, 2016;Baron, 2004;Golden, 1987;Korkman, 1998;Korkman, Kirk, & Kemp, 2007;Reitan, 1959;Reitan & Wolfson, 1985;Tramontana & Hooper, 1988;Weiler, Willis, & Kennedy, 2019).
How were the qualitative and quantitative approaches combined in the research by T.V. Akhutina and her colleagues? If, as in making a qualitative diagnosis of adults, one can proceed from a general idea of the normative performance of tasks and take it as 0, then when working with children, it is necessary to quantitatively assess the qualitative characteristics of task performance by typically developing children of different ages. To capture the qualitative speci city of the test performance and be able to assess it quantitatively, it has been necessary to standardize the presentation and analysis of the tests, and to develop parameters for evaluating not only the productivity of test execution but also speci c errors which re ect both primary and secondary di culties in completing the tasks. e division of errors into primary and secondary types is a distinctive feature of neuropsychological analysis in the Vygotsky-Luria approach. Combining the productivity parameter for a certain test with the primary errors during its performance, and then summing them up with the parameters of other tests aimed at detecting the same primary defect, has led to the development of integral estimates, or indices (Korneev & Akhutina, 2016). e method of adding the parameters of several tests has also been proposed within the quantitative approach; it is known as latent process analysis (Miyake, Emerson, & Friedman, 2000). To solve the problem of analyzing the component comprising executive functions, Miyake and his colleagues concentrated on "the task impurity problem. " Due to the complexity of any human activity, there are no tasks that are unambiguously associated with only one cognitive function without involving others. Latent process analysis makes it possible to reduce the impurity of tests. In discussing executive func-tions, the authors write: "A latent variable is essentially a hypothetical construct created by statistically 'extracting' what is common among the multiple tasks chosen to tap that construct. In the case of our study, we created three latent variables that corresponded to the Shi ing, Updating, and Inhibition factors, respectively. It is important to point out that these latent variables were 'purer' measures of the target executive functions because each latent variable contained only what was shared among all three tasks and not what was speci c to each task. " (Miyake et al., 2000, p. 178) A similar method of adding unidirectional parameters to aggregate a set of results which would give a "purer" objective estimation has also been developed in child neuropsychology based on the Vygotsky-Luria theory. It is applied to assess not only executive functions but information-processing functions and functions of activation. is approach has evolved from the principles and methods of diagnosing local brain lesions in adults described by Vygotsky and Luria (Luria, 1965(Luria, , 1976(Luria, , 1980. It has turned out to be e cient even in solving such di cult issues in child neuropsychology as the analysis of cognitive functions for the selection of e ective methods to overcome learning di culties (Akhutina & Pylaeva, 2012).
In the context of the assessment of typical and deviant development, a further elaboration of optimal methods for evaluating cognitive functions in terms of efciency, conciseness, and ecological validity turns out to be an important issue. In particular, we should examine a set of quantitative parameters of the quality of test performance, which we can use when calculating integral estimates or indicators of various cognitive functions. e need for such veri cation grows out of the problem of the "insu cient purity" of any test or task. To check the correctness of distributing the parameters into indices, we use the method of con rmatory factor analysis.
Within our work, a comprehensive neuropsychological examination for 6-to-9 year-old children was employed to diagnose the children's state of higher mental functions (Akhutina, 2016). is battery consisted of 20 tests; their performance was assessed according to numerous parameters that are used to determine a neuropsychological conclusion about the state of a subject's cognitive functions. e system for calculating integral indices has also been developed to determine generalized quantitative indicators of the state of di erent cognitive processes (Korneev & Akhutina, 2016). Each of those indices consists of a set of performance indicators for di erent tests within the examination.
At present, our approach applies a set of indices that allows for evaluating the following processes: programming and control of voluntary actions (executive functions); serial organization of movements and speech; processing of kinesthetic information; processing of auditory information; processing of visual information; processing of visual-spatial information; hyperactivity/impulsivity; and fatigue/slow tempo. When diagnosing the state of cognitive functions in preschool and primary school children, that set of indices is quite complete and makes it possible to assess all of the most important cognitive components that intensively develop at that age, and are important for the child's success in learning (Ardila & Rosselli, 1994;Klenberg, Korkman, & Lahti-Nuuttila, 2001;Korkman, Kemp, & Kirk, 2001;Stiles, Akshoomo , & Haist, 2013;Vanvooren, Poelmans, De Vos, Ghesquière, & Wouters, 2017). It is important that the indices not indicate global multifactorial constructs like memory or language but rather the component processes that underlie complex cognitive functions in the children.
In recent years, computerized batteries for neuropsychological examinations have been developing rapidly. e most famous of them is CANTAB (Luciana & Nelson, 2002), but there are also several others, including MINDS (Brand, & Houx, 1992;Brand, von Borries, & Bulten, 2010), ANAM, ImPACT, CogState, and CNS-VS (see Parsons, 2016 for a review). We also have developed and used a battery of tests aimed at evaluating executive functions, functions of activation (arousal), and functions for processing visual-spatial and auditory information (Korneev et al., 2018). Our computer battery is based on ideas of the Lurian assessment and provides a more objective and standardized way to estimate cognitive functions in children. It can be used as a method for screening children with a risk of learning or other disabilities. In this battery we apply the same approach when certain cognitive functions are evaluated by integral indices that consist of a set of indicators derived from the performance of several tests. A separate task may be the creation of such integral indicators that combine the performance parameters of both face-to-face and computer-based neuropsychological tests. e techniques for calculating the indices are described in our previous works (Korneev, & Akhutina, 2016;Akhutina, Korneev, Matveeva, Gusev, & Kremlev, 2019). A detailed description of the procedure is beyond the scope of this paper, but in brief, the main idea is to standardize the parameters (productivities and speci c mistakes in various tests) and to sum up the z-values into integral estimations.
Since the elaboration of the indices is based on theoretical analysis and the experience of neuropsychological examinations, suggestions about the composition of the indices should be veri ed in targeted studies. is current work attempts to verify the structural validity of a set of indices of di erent cognitive functions that we developed speci cally for preschool and primary school children. Con rmatory factor analysis is one of the methods of that assessment. is method is quite common in the study of the structure of cognitive functions. us, there are studies of the factor structure of executive functions that allow for identifying and assessing their di erent components: inhibition, updating, and shi ing (Miyake, 2000;Friedman & Miyake, 2017). is approach is also employed in the analysis of the structure of intelligence based on the results of the Wechsler Intelligence Scale for Children, which distinguishes four factors: verbal, perceptual, processing speed, and working memory abilities (Bodin, Pardini, Burns, & Stevens, 2009). Using con rmatory factor analysis, we intend to assess the structural validity of the indices that characterize the state of di erent cognitive functions in children, while including various indicators of task performance in both paper and pencil and computerized neuropsychological examinations. Our research questions may be formulated in the following way: 1. To what extent can the composition of neuropsychological indices, which were developed earlier and based on the theory and practice of child neuropsychology, be verified on a large sample of typically developing children ages 6 to 9 years old? 2. Is it possible to combine the results of a face-to-face examination and computer tests for a more accurate and reliable assessment of the state of cognitive functions?

Method
Participants is study involved 299 children ages 5.1 to 9.5 years. Of these, 111 were preschoolers (56 boys and 55 girls; average age 6.45 years); 82 were rst graders (31 boys, 51 girls; average age 7.67 years); and 106 were second graders (62 boys, 44 girls; average age 8.58 years). e participants were from regular Moscow kindergartens and primary schools; their families had middle socioeconomic status. None of the participants had any diagnosed neurological or developmental disorders. e parents of all children gave informed consent for them to participate in the study.

Measures
Face-to-face neuropsychological examination e neuropsychological examination adapted for children ages 6 to 9 years old was used to determine the state of their cognitive functions (Akhutina, 2016). Although a comprehensive examination includes 20 behavioral tests, in this study we analyzed data from only some of them. e list of the tests with a short description of the procedure is presented in Table 1. Table 1 Test battery for the neuropsychological assessment of children 6-to-9 year-olds (partial)

Go/No Go Task, Reciprocal Motor
Program Test e test consists of two series: 1. One knock should be responded to with two knocks, and two knocks with one knock; 2. One knock should be responded to with two knocks, and two knocks with no knock.

Verbal Fluency
Tests e test consists of three series. e participant must name as many words as he can in one minute: 1. any words; 2. names of actions; 3. names of plants.

Odd one out
Five series of ve words are presented aurally. e child has to nd the odd one and explain the choice.
Counting e participant must count from 1 to 10; from 10 to 1; from 3 to 7; from 8 to 4; or count from 20 subtracting 3 at a time.
3 Positions Test, or "Fist-Edge-Palm" e participant must remember and automate a series of hand-movements: e participant is asked to reverse the con gurations of his or her hands repeatedly and simultaneously from palm to st, so that when the st is opened in one hand, it is closed in the other.

Kinesthetic information processing
Finger Position Test e test consists of three series: 1. imitation of nger positions ( ve positions for each hand); 2. reproduction of nger poses using proprioceptive memory (three positions for each hand); 3. transferring nger poses from one hand to the other (three positions for each hand).
Oral Praxis e participant must perform movements and poses using the orofacial muscles on verbal command (ten tasks).

Auditory information processing
Verbal Memory Test e participant must remember two groups of three words each. ree attempts are given, and delayed replay is also evaluated.

Visual information processing
Visuo-perceptual Tests e participant must identify superimposed, crossed out, and un nished images (22 pictures).
Design Fluency Tests (DFluency) e participant must draw any eight objects and any eight plants.

Visual-spatial information processing
Visual-Spatial Memory Four nonverbal gures are presented for eight seconds. e participant must remember and draw them. ree attempts are given, and delayed replay is also evaluated.
ree-Dimensional Drawings e child must copy a three-dimensional picture of a house.
Test performance was evaluated by an expert neuropsychologist according to several parameters that indicate the child's ability to understand the task, the accuracy and e ciency of performance, and speci c mistakes revealing any weakness of the tested function (primary mistakes) and other functions (secondary mistakes). We calculated productivity as the number of correct answers, marks of accuracy in ordinal scales (from 0 to 3), and numbers of speci c or unspeci c mistakes. For instance, when analyzing the Finger Position test, kinesthetic di culties (speci cally the prolonged search for a pose) were assessed as primary errors and impulsive performance as a secondary error.
According to the results of the examination, the neuropsychologist could evaluate the state of the functions of activation: symptoms of fatigue, sluggishness, a tendency to perseveration, impulsivity, and hyperactivity (on an ordinal scale from 0 to 3).

Computerized neuropsychological tests
In this study, we used tests from the battery of the Computerized Neuropsychological Assessment in 6-9 Year-old Children (Korneev, Akhutina, Gusev, Kremlev, & Matveeva, 2018). e battery consists of 10 tests; ve of them were used in the present work: 1. e Dots test (Davidson, Amso, Anderson, & Diamond, 2006). e test consists of three subtests; each of them involves 20 stimuli. In the rst subtest, the stimuli (hearts) are presented on the computer screen, in a quasi-random order, to the le or to the right of the screen center. e child's task is to press the button on the side where the stimulus appears as quickly as possible. e subtest assesses the ability of the participant to follow instructions and the speed of a simple motor reaction. e second subtest evaluates the child's ability to inhibit the "natural" response that is irrelevant to the task: another stimulus (a ower) appears on the screen; the task is to press the button as quickly as possible on the side opposite to the one where the stimulus appears. e third subtest evaluates the child's ability to switch between the two parallel programs: two types of stimuli (hearts and owers) are presented alternately on the screen; the task is to press the key on the same side where the heart appears, and on the opposite side where the ower appears. is test is assessed according to the average response time and productivity (the number of correct responses).
2. A computer version of the "Schulte Tables" (Korneev et al., 2018). e test consists of ve subtests; each presents a table consisting of 20 cells on a touch screen. In those cells, there are two series of numbers from 1 to 10 arranged in quasi-random order; one series consists of black numbers, while the second set of numbers is red. e participant must search for and indicate the numbers in a certain order by touching the screen with a nger. e rst subtest calls for pointing to the black numbers from 1 to 10, followed by the red numbers from 1 to 10 in the second subtest; then in the third subtest, the black numbers from 10 to 1; in the fourth subtest, there should be two parallel series showing red and black numbers in ascending order (1 black, 1 red, 2 black, 2 red, etc.), and in the h, the participant must indicate red numbers from 10 to 1. Such a set of tasks makes it possible to assess the children's ability to master a simple action program (the rst and second subtests), a more complicated reverse program (the third and h subtests), and the most di cult, a "parallel" action program (the fourth subtest). ey have to switch their attention from one program to another and must inhibit inadequate responses. Based on the results, we calculated the average time of searching for a number as well as the error count, both for the whole test and in the ve subtests separately.
3. e Cancellation test for preschoolers and primary school children (Korneev et al., 2018). e test consists of three subtests. e touch screen displays a table consisting of six similar elements (geometric gures in the version for preschoolers, letters in the version for young students). In the rst subtest, the child's task is to nd and mark one of the stimuli, in the second one they nd and mark the other one, and in the third subtest the target is both of those stimuli. us, during the rst two trials, we assess the child's ability to keep their attention on a simple task for quite a long time, and in the third trial, the ability to switch to a more complicated instruction is assessed. e evaluated parameters are tempo (the number of correct answers per minute) and accuracy (the percentage of correct answers).
4. e Corsi Tapping Block test (Milner, 1971; a computerized version for children by Korneev et al., 2018). Nine cubes presented on the touch screen light up one by one. e task is to remember their positions on the screen, and a er their presentation, the participant must reproduce the sequence of the highlighted cubes. e trial starts with a row of two cubes; with every right answer, the length of the row increases. e indicators in this trial are the maximum length of a correctly reproduced sequence, the average time of the rst response, and the average time of pauses within the sequence.
5. e Understanding of Similar Sounding Words test (USSW; Korneev et al., 2018). e child is presented with a set of 10 pictures of distinct objects whose names di er in one sound; for instance, "bochka-pochka" (barrel-bud). en a sequence of words is presented aurally (a total of eight sequences, each two to ve words in length). e child must indicate the corresponding pictures in the same order. We evaluate the percentage of correctly reproduced words (relative to the total number of responses) and the numbers of di erent mistakes (substitutions of similar and dissimilar sounding words and omissions).

Analysis
Con rmatory factor analysis was used to test our hypothesis about the possibility of identifying the parameters that characterize di erent groups of cognitive functions. e parameters of the tests' performance (productivity, speci c errors, and reaction time) were used as the exogenous variables (indicators) in the model, and the cognitive functions were included as endogenous variables (factors). Since some of the performance indices of the neuropsychological examination were estimated on ordinal scales, we used the method of weighted least squares with the means and variation adjusted (WLSMV; Muthén & Muthén, 2012). is method is applied in the case of ordinal scales and is resistant to a non-normal distribution of data. e analysis was conducted in R version 3.6.0 with the Lavaan package (ver. 0.6-9, Rosseel, 2012).
To assess the quality of the models, we used the following rules: for CFI and TLI, values higher than .90 re ect a good model t; for RMSEA, less than .08 indicates close t (Schumacker, & Lomax, 2010).

Models with parameters of face-to-face assessment
Several models were constructed and tested. e rst model corresponded to the composition of indices used in practice (Korneev & Akhutina, 2016 5) Visual information processing (Vis): productivity, number of visual errors in the Visual Perception test, number of well-recognized pictures in the Design Fluency tests, and the tree drawing score in 3-Dimensional Drawing; 6) Visual-spatial information processing (VSp): productivity of the first and third reproduction of stimuli in the Visual-Spatial Memory test, number of transformations of stimuli into a sign, severity of weakness of the right-hemisphere or left-hemisphere strategy in 3-Dimensional Drawing, and number of spatial mistakes in the Finger Positions test; 7) Sluggish tempo (ST): indices of fatigue, a lower tempo of task performance, and severity of the tendency to perseveration; and 8) Hyperactivity/Impulsivity (HImp): indices of hyperactivity and impulsivity. e model also allowed for correlations between all those factors. e estimates of this model turned out to be acceptable but not too high (χ 2 (915) = 2295.921, CFI = 0.907, TLI = 0.899, RMSEA = 0.071); the full data on the coe cients of the model are given in the Appendix, Table 1A. However, since some variables in this model had very small factor loads, we modi ed the model to exclude such cases. On the EF factor index, the index of mastering in the 3 Positions test was excluded. On the Kinest factor, performance in the Oral Praxis test was excluded (this index displays the ceiling e ect). On the Aud factor, we excluded the vowel change in the Verbal Memory test (this is a rare error) and verbal errors in the Visual Perception test. On the Visual-spatial factor, the number of transformations into a sign in the Visual-spatial Memory test and spatial errors in the Finger Positions test were excluded.
Meanwhile, the productivity of the rst repetition and the number of distortions during the reproduction of words in the Verbal Memory test were added to the factor of processing the auditory information. e estimates of this modi cation improved (χ 2 (750) = 1692.926, CFI = 0.918, TLI = 0.910, RMSEA = 0.065; the full data about the coe cients of the model are given in the Appendix, Table 2A), and all of the factor loads were signi cantly di erent from zero at the level of p < 0.05.

Model with parameters of both face-to-face and computerized assessment
e addition of the indices of computer test performance into the model was the next step. e following were added: 1) to factor EF: productivity of the third subtest of Dots, number of mistakes in the fourth Schulte Table, and the total accuracy of the Cancellation test; 2) to factor Aud: productivity and similar replacements in the USSW; 3) to factor VSp: productivity in the Corsi Test and response time (search) in the fourth Schulte Table; 4) to the factors of sluggishness and hyperactivity: response times in the first subtests of Dots and Schulte Table, the average interval between responses in the Corsi Test, and the average tempo in the Cancellation test.
e estimates of that model improved (χ2(1241) = 2579.507, CFI = 0.935, TLI = 0.930, RMSEA = 0.060); the full data on the model coe cients are given in Table 2. e correlations between the factors are shown in Table 3.

Discussion
A con rmatory factor analysis of the performance parameters of the various tests from the batteries of face-to-face and computer neuropsychological examination made it possible to identify the factor structure that corresponded to the proposed structure of integral estimates of di erent groups of cognitive functions. We detected and con rmed two factors associated with the functions of activation at the level of empirical data. ese were the hyperactivity/impulsivity factor, which correlated with disturbances in the form of ADHD (Barkley, 1998), and the factor of sluggishness, which manifested in the syndrome of a sluggish cognitive tempo (Becker, Marshall, & McBurnett, 2014;Becker & Willcutt, 2019). ese results correspond to the data obtained in fMRI research (Fassbender, Kra , & Schweitzer, 2015). It is worth emphasizing that, in our model, the same indices of performance times for the computer tests had signi cant loads in both factors but with an opposite sign. is also justi es separating the neurodynamic functions in that way.
As to the functions associated with information processing, we received con rmation at the level of the structural model for the validity of the division of individual factors for the processing of kinesthetic, auditory, visual, and visual-spatial information. e possibility of identifying modally speci c mechanisms of information processing in solving di erent tasks is under discussion in the literature. ere are arguments both in favor of (Barsalou, Simmons, Barbey, & Wilson, 2003) and against (Anderson, Qin, Jung, & Carter, 2007) that division. Within the framework of our approach, the results obtained on a sample of typically developing children pointed toward such a division, at least at the level of behavioral indicators of neuropsychological task performance.
As to the distinction of factors connected with voluntary activity, following A.R. Luria (1976Luria ( , 1980 in this case, our model distinguished the factor of the programming and control of activity (≈ executive functions) and the factor of the serial organization of actions. e results of the con rmatory analysis also rea rmed that division. Executive functions are important for the performance of almost any voluntary activity; some studies show that they have a heterogeneous structure and can be divided into separate groups of functions (inhibiting, updating, and shi ing; see Miyake et al., 2000;Friedman & Miyake, 2017). Such detailing was not carried out in our research; it may require a separate study based on neuropsychological examination.
It is noteworthy that the identi ed factors highly correlate with each other. is is not surprising, as it is di cult to expect them to be independent of each other. Solving even a simple task inevitably engages several groups of functions. is corresponds to ideas about the relationships between functions in neuropsychological theory. We attempted to identify the indices of the states of certain speci c functions, and we managed to do that to some extent, although at the same time we detected quite a close relationship between functions. Whether it is possible to identify more "pure" indicators of certain functions within the framework of ecologically valid tasks in the examination is a topic for further discussion. Such a "targeted" assessment is possible through more speci c laboratory studies held in the framework of experimental psychology.
Another signi cant result obtained in our study was the preservation of, and even some improvement in, the quality of the model by including the results of the computer tests. is indicates that the combined usage of face-to-face and computerized neuropsychological examinations can improve the reliability and accuracy of the assessment of the states of di erent functions. ere are data about a weak correlation between the results from the computer tests and the traditional neuropsychological examination (Smith, Need, Cirulli, Chiba-Falek, & Attix, 2013). Our model detected that the same indices (reaction time) may be associated with di erent factors with an opposite sign. is is explainable and meaningful, but it can also be the reason for the weakening of simple linear relationships between the results of di erent methods. Constructing and testing models like the one described in our paper can be an efcient way to investigate the consistency of the tests. us, the results of our con rmatory analysis support the structural validity of identifying our hypothesized groups of functions and the validity of evaluating them by indices that include indicators of the performance of di erent tasks within the neuropsychological examination. is approach makes it possible to carry out a detailed and comprehensive analysis of the cognitive states of children at preschool and primary-school ages, with typical development and with di erent disorders. In the typical samples, this approach makes it possible to distinguish children with a relative weakness (de ciency) of certain abilities associated with the uneven development of some functions and, if necessary, to arrange a preventive remedial intervention. With more pronounced behavioral problems classi ed as disorders, such an assessment provides an opportunity to analyze the structure of the defect and to suggest the most e cient ways to correct it.

Conclusions
e present study proposed and evaluated several models describing the possible composition of indices of di erent cognitive functions, made up of indicators of performance on neuropsychological examinations by preschool and primary school children. ey include the following groups of cognitive functions: programming and control of voluntary actions (executive functions), serial organization of movements and speech; the processing of kinesthetic information; the processing of auditory information; the processing of visual information; the processing of visual-spatial information; hyperactivity/impulsivity; and fatigue/slow tempo. e nal model showed good consistency with the data. is con rmed the structural validity of the proposed scheme for quantifying the state of the cognitive sphere in children.
Including the performance indices from two methods (face-to-face and computerized tests) had the important result of improving the quality of the model. Furthermore, this model may be useful in the research of the patterns of cognitive capacities in children with typical and deviant developments. Secondly, our ndings make it possible to construct more detailed models that clarify the structure of cognitive functions, especially executive functions, in children. In general, our work shows that the qualitative approach to neuropsychological diagnostics, developed by A.R. Luria (Luria, 1980), can be e ectively combined with quantitative analysis of neuropsychological data.

Limitations
Our study had some limitations. We did not analyze the in uence of socioeconomic factors, although they may be important. We selected schools with approximately the same average socioeconomic level of children, but more detailed and precise analysis may be needed in future studies. Our sample included only children of the metropolis, so it would be important to test our ndings on samples from other regions.
Co mputer methods allow us to see the quantitative characteristics of some cognitive functions, but they still do not replace expert qualitative assessments. Computer tests can be used as a screening method: to identify children with risk for learning disabilities. e quantitative estimates obtained with their help can complement the qualitative assessment of the expert.
We have developed and discussed the integrative indices that can be useful in the situation of screening or for the generalization of results of the assessment in large samples. But this approach is less sensitive than a neuropsychological conclusion made by an expert. We have to remember that both the qualitative and quantitative approaches have both strengths and weaknesses.

Ethics Statement
Parents of all subjects gave their informed consent for participation of their children in the study. e study received ethical approval from the Ethical Committee of the Faculty of Psychology, Lomonosov Moscow State University (Moscow, Russia).

Informed Consent from the Participants' Legal
Written informed consent to participate in this study was provided by the participants' legal guardian/next of kin.

Author Contributions
Tatiana Akhutina and Aleksei Korneev developed the concept of the study and performed the theoretical analysis. Ekaterina Matveeva collected the data and prepared it for analysis. Aleksei Korneev performed the computations. All authors prepared the original dra , carried out the review and editing of the manuscript, and contributed to the nal version.

Con ict of Interest
e authors declare no con ict of interest.