Executive functions in children with dyslexia

ABSTRACT This study aimed to verify whether children with dyslexia have difficulties in executive functions (shifting, working memory, inhibition). Methods: A sample of 47 children (ages 8-13 years) participated in the study: 24 who were dyslexic and 23 controls with typical development. A battery of neuropsychological tests was used. Results: Results revealed executive function difficulties among the dyslexic children when compared with controls, encompassing selective attention modulation processes, shifting, and inhibitory control. These difficulties appeared to be affected by phonological working memory deficits, typically associated with dyslexia. Conclusion: Our findings support the consensus among scholars regarding the central involvement of phonological skill dysfunctions in dyslexia.

regard, the study emphasized the importance of a more diversified use of quantitative and qualitative procedures to assess executive functions in clinical groups, particularly in dyslexic patients.
The Wisconsin Card Sorting Test (WCST) a neuropsychological test that allows assessment of mental flexibility and inhibitory control skills, has been adopted in dyslexia studies, among other standard executive function measures. In this procedure, participants are asked to identify different categorical criteria for the association of cards, following verbal feedback. In one study 17 , it was observed that dyslexic children identified a smaller number of categories compared with control groups, in addition to showing greater attentional engagement difficulties. These findings suggest problems in shifting, inhibition, and self-monitoring skills. In another study 18 , when compared with children fluent in reading, dyslexic children required more cards to complete a category, completed fewer categories and made more errors, which relates to shifting difficulties. A study published with Brazilian Portuguese speakers illustrated the difference between dyslexic children and controls regarding the number of categories completed and the number of cards used in the WCST; the dyslexic children completed a smaller number of categories and required a higher number of cards to complete the task, potentially indicating mental flexibility and problem-solving impairments 19 .
A performance evaluation of children with and without dyslexia in the Conners' Continuous Performance Test (CCPT) showed that dyslexic children had more flaws in inhibitory control (commission errors), higher response variability, more perseverative errors, and lower response consistency over time. These results point to difficulties regarding inhibitory control, including impulsivity, and sustained attention problems 20 .
More recently, deficits were identified in several executive function domains in dyslexic children, such as verbal fluency, attention, and phonological working memory, in addition to those commonly found in phonological awareness tasks 21 . In turn, difficulties were identified in verbal fluency and shifting among Portuguese children with dyslexia 4 .
On the other hand, some studies failed to reach consistent results regarding the presence of broad executive function disabilities in reading and writing disorders. Dyslexic individuals were submitted to executive functioning skills (inhibition, working memory and attention), phonological skills, and time perception measures. The results showed impairments only in phonological skill measures 22 .
In a study conducted with Brazilian Portuguese speakers, children aged six to eight, referred by their teachers as good or poor readers at the beginning of the literacy teaching process, were compared for selective attention, inhibition, flexibility, and working memory skills tests. Only working memory and flexibility impairments were directly related to reading difficulties 23 . Therefore, considering the controversy surrounding the involvement of broad executive function disorders in dyslexia, the purpose of this study was to ascertain whether dyslexic children have difficulties in executive function and in which of its key processes they occur: shifting, working memory and response inhibition.

Participants
A sample of 47 children between the ages of eight and 13 participated in the study: 24 dyslexic children (dyslexia group -DG) with no comorbidities; and 23 children with typical development (control group -CG), paired by age, sex and school attendance (private or public school).
Participants of the DG were selected from a sample of 110 children assessed at a learning disabilities outpatient clinic in Brazil (Núcleo de Atendimento Neuropsicológico Infantil Interdisciplinar). Participants from the CG were assessed at their schools, and were required to show adequate academic performance based on the respective school grade and age. The period of recruitment extended from 2014 to 2015. All the children were submitted to a broad neuropsychological battery comprising tests and tasks involving oral language, reading and writing, in individual sessions of approximately one hour each. The dyslexia diagnosis was based on the following instruments: CONFIAS -the Phonological Awareness, Instrument of Sequential Assessment; the TDE School Performance Test; reading and writing of words and pseudowords; reading silently and reading aloud task; and a reading comprehension task (see Cruz-Rodrigues et al. 19 and Barbosa et al. 24 for further explanations about these tests).
The dyslexic children were previously diagnosed based on the following criteria, as established in the DSM-5: persistent reading and writing difficulties from the beginning of the literacy teaching process; selective deficits in reading and writing accuracy; and slow reading. All the children of the CG had good academic performance and attended public or private schools.
Neither group included children with an intellectual performance below 90, as assessed by the Wechsler Intelligence Scale, sensory or motor disabilities, or a history of psychiatric/neurological comorbidities such as attention deficit hyperactivity disorder, or anxiety disorder as evaluated by the Brazilian version of the Child Behavioral Checklist and SNAP-IV Parent Rating Scale. This information was obtained from the respective parents or legal guardians, who signed an informed consent form authorizing the children's participation in the study, which was approved by CEP/UNIFESP.

Procedures
The battery used for the assessment of executive function included the following tests:

Attention and Inhibition
Conners' Continuous Performance Test -software for Windows (CCPT 4) 25 -presented on a laptop, covering an application time of approximately 20 minutes, including training and testing activities. Participants are required to press the keyboard space bar when any letter appears on the screen, except for "X". Each letter is shown for approximately 250 milliseconds. There are 324 target stimuli and 36 non-target stimuli ("X" letters). Six stimuli blocks are shown, each with three sub-blocks of 20 items each (letters shown) and interstimulus intervals of one, two and four seconds 26,27 .

Phonological Working Memory
Digit Span (WISC-III 28 ), forward and backward order. In this classic phonological working memory task, the examiner reads progressively larger number sequences aloud (up to eight numbers). For each forward order sequence, the child repeats the numbers in the same order they were given. For each backward order sequence, the child repeats the numbers in reverse order. Each item has two attempts, with different numbers, and each attempt has the same number of digits 28 . The final score represents all numbers correctly repeated in sequence ( forward and backward).
Brazilian Children's Test of Pseudoword Repetition 29 . The task comprises 40 pseudowords. The examiner places a blindfold on the children to prevent them from identifying any facial queues and asks them to repeat pseudowords exactly as they were heard. One point is given for each correct answer. This task assesses phonological working memory.

Visuospatial Working Memory
Corsi Block-tapping test -forward and backward order 30 . This test comprises nine blue blocks irregularly arranged over a white board. During the task, the examiner taps predefined and progressively longer block sequences, and the participant must repeat the movements in the same sequence. In the backward test, the participant is required to repeat the block-tapping sequences in reverse order. There are two attempts for each sequence. The test is interrupted when the participant makes two consecutive mistakes or completes the nine-digit sequence.

Shifting
Wisconsin Card Sorting Test -manual version. This test consists of a set of 128 stimulus cards with different patterns on white backgrounds, as well as four target cards placed in front of the participant, featuring assorted attributes (shape, number, and color). The target cards, in order of presentation, are: one red triangle, two green stars, three yellow crosses, and four blue circles. The participants are required to match the cards from the set with the target cards, one at a time, based on the respective attributes. There is a predefined matching sequence (color, shape, quantity), which is not revealed to the participants. In other words, the strategy initially required from the participants is to match cards based on color. After correctly matching ten cards, they must then match cards based on the shape. After ten cards are matched, they must then match cards based on the quantity, and so on successively until all six categories are completed. The examiner informs the participants whether the adopted strategy is correct or incorrect 31,32 .

Statistical analysis
The statistical analysis was performed using the SPSS statistical package -Version 18. The level of significance adopted was 5%. The Chi-square test was performed for string variables. Levene's test was initially performed for numeric variables, in order to analyze distribution. Since the variables had normal distribution, the Student's t-test was performed. However, we identified inconsistencies in the level of intelligence of children between the DG and CG, which was higher in the latter. Therefore, we performed the Analysis of Covariance (ANCOVA) for significant variables in the Student's t-test.

RESULTS
The sample characterization is described in Table 1. The distribution of groups based on age, sex, and school type was not divergent due to the pairing strategy adopted.
In the CCPT (Table 2), the measures that showed disparities were the number of omissions (p = 0.01) and percentage of omissions (p = 0.02), which were higher in the DG compared with the CG; and the reaction time (Hit RT) (p = 0.001), in which the DG was significantly slower than the CG (more time).
Regarding working memory, the DG underperformed in the Brazilian Children's Test of Pseudoword Repetition test for five-syllable items (p = 0.0005) and the backward Digit Span test (p = 0.007). No differences were observed between both groups in the Corsi Block-tapping test (Table 3).
In the WCST (Table 4), the DG underperformed only in the percentage of perseverative errors (p = 0.04). It is important to note that other WCST variables with significant

DISCUSSION
The results obtained in this study indicated significant discrepancies in the performance of the dyslexic children surveyed in some of the executive function measures, in comparison with children with typical development, thus reinforcing current evidence that reading and writing disabilities are at least partly associated with executive function disabilities.
Regarding the inhibition domain, as inferred by the CCPT test results, a slow motor response and a tendency towards omission-type errors were identified, whereas no impulsive response inhibition difficulties were encountered, although other studies had identified inhibitory control (impulsivity) and sustained attention difficulties in their studied sample 20 .
Another study based on CCPT applied to dyslexic children did not reveal any differences in surveyed measures, when compared with the control group 30 . Therefore, our results seem to support the notion that despite not being clinically diagnosed with attention deficit/hyperactivity disorder, dyslexic children tend to show difficulties in modulating attention, rather than inhibitory control disabilities 18 .
Regarding the working memory domain, as expected, phonological working memory specifically appeared to be the most fragile cognitive area among the participants. In fact, no significant differences were identified between the children in the DG and the CG in nonverbal working memory measures (assessed via the Corsi Block-tapping test). This result indicates that dyslexia is not associated with global working memory losses, but rather with specific deficits in the processing of verbal materials (phonological working memory), which has been described in several studies 11,12,13 . Furthermore, the ability to retain phonological working memory was also affected, as assessed via the Brazilian Children's Test of Pseudoword Repetition test that indicated difficulties centered in phonological skills, rather than in working memory manipulation skills, which would be more intimately related to executive functions.
A study by Engel et al. 33 confirmed that working memory measures were not influenced by the environmental variables of studied children, as opposed to analyses regarding the level of verbal intelligence 33 . Therefore, it is possible to infer, that although working memory and intelligence are intimately related, working memory performance is a better indicator of a child's learning potential than intelligence measures, which mainly assess prior knowledge acquired at home and at school. Working memory measures are, therefore, seen as far more useful procedures for learning disabilities diagnostic processes.
Finally, considering the results of the WCST, a higher percentage of perseverative errors were recorded in children of the DG, compared with the CG, suggesting that shifting and inhibition abilities may be compromised in dyslexic patients 17 . No differences regarding the number of identified categories were encountered, as opposed to other findings 19 .
On the other hand, problems in the phonological working memory may also potentially explain the increased number of perseverative errors within the DG. That is, although the WCST is traditionally used as a shifting measure, it is a complex task that also requires phonological working memory abilities to maintain the current memory category until the task is completed. In fact, recurring evidence indicates that phonological working memory deficits are associated with a higher frequency of perseverative errors in the WCST 6,34,35 . These deficits supposedly affect the permanence of the current category in the memory and the benefit of the examiner's feedback, leading the child to commit more errors of this type 3 . Therefore, despite some authors employing the WCST in children with learning disabilities due to its nonverbal nature, this does not mean that the test does not involve verbal skills, such as the phonological working memory 16,18 .
In summary, this study observed executive function difficulties among dyslexic children, encompassing selective attention modulation processes (translated into slower responses and a tendency towards omission errors in the CCPT), as well as shifting and inhibitory control ( frequency of perseverative errors in the WCST), which seemed to be affected by phonological working memory deficits, typically associated with this diagnosis. Limitations of the study include the small sample size and the use of some instruments not fully adapted for the Brazilian population, although we used a control group for comparison. Nevertheless, our findings support the general consensus among scholars regarding the central involvement of phonological working memory alterations in dyslexia, which better explain the inherent difficulties with executive functions in this diagnosis. Further research on this matter is required, primarily to assess the influence of phonological working memory on executive functions.