Our search produced 5746 papers. These were scanned and 5650 papers were excluded as they did not fulfil the inclusion criteria for this review. Ninety-six articles were retrieved in full, and of these, 39 were excluded (participants in five papers were either children younger than school age or adults; relevant data were not reported in 27 papers, six papers turned out to be reviews or commentaries, and one paper was a case study). A total of 57 papers that satisfied our selection criteria were identified.
Visuospatial deficits are perhaps the best-established characteristic of the cognitive profile of NF1 in childhood. The first detailed study describing learning deficits in NF1, by Eliason,37 found that the majority of children with NF1 suffered from visuoperceptual (non-verbal learning problems), although this label did not comprehensively explain all aspects of the functioning in these children. A number of studies have investigated this further, using a variety of measures that focus on the perception of angulation (e.g. the JLO test), visual organization (e.g. the Hooper Visual Organization Test), and object recognition (e.g. Benton Visual Form Discrimination Test and the Birmingham Object Recognition Battery). Children with NF1 have been found to perform significantly more poorly on visuoperceptual tasks than sibling comparisons,3,15–18,22–24,36,38–41 as well as scoring at least 1SD below standardized norms.27,42 There are some conflicting findings though: Clements-Stephens et al.33 found that, although children with NF1 scored significantly more poorly than comparison children on the JLO test and on two perceptual subtests of the Developmental Test of Visual Perception, there were no significant differences between the groups on the Hooper Visual Organization Test; Descheemaeker et al.35 failed to show a substantial deficit in visuomotor skills alone; and De Winter et al.43 found that children with NF1 demonstrated larger deficits on pure visuoperceptual tasks (the JLO test, recognition–discrimination task) than tasks requiring visuomotor components as well. However, in the studies of both Clements-Stephens et al.33 and Descheemaeker et al.35 sample size was fairly small, so the failure to find a statistically significant difference may be due to inadequate power in these studies. Brewer et al.,32 using hierarchical agglomerative clustering analysis, found that 14% of the children with learning problems in their sample fell into what they called the ‘visuospatial constructionally deficient group’, having special problems in the visuospatial constructional and fine motor coordination domains. This is a surprisingly small proportion, suggesting that it is not very common for visuospatial problems to be the only NF1-related cognitive problems of a child with NF1.
It seems that the JLO test captures particularly well the type of visuospatial impairment common in NF1; for example, Hyman et al.3 found that 56% of children with NF1 scored at least 1SD lower than comparison children, and Krab et al.17 and Moore et al.24 also found sizable deficits in the performance of children with NF1 on this particular task. Schrimsher et al.39 administered several different visuospatial tests to children with NF1, as well as comparison children. They then performed a discriminant function analysis and found the JLO test to be the best predictor of group membership. However, as pointed out by Levine,7 it is possible that the JLO test taps into other aspects of cognitive performance, in addition to visuoperceptual function, that makes it sensitive to deficits in NF1, such as executive function.18
Motor control and coordination, visuomotor function
Motor performance is closely related to visuospatial function, as pointed out, for example, by Levine et al.7 Parent reports of poor coordination and clumsiness are common in clinical practice, and Krab et al.17 found that over 40% of children receive remedial teaching to improve their motor performance at school, whereas in the study by Hyman et al.,3 53% of children with NF1 were found to have poor handwriting, according to parental reports (in contrast to 6% of children in a comparison group). Motor performance has been measured by tasks such as the Grooved Pegboard, the Beery Visual–Motor Integration Test, the Rey Complex Figure test, subtests from the Amsterdam Neuropsychological Tasks battery, the prism task, and the Bruininks–Oseretsky Test. Children with NF1 appear to struggle with both their fine and gross motor skills.3,16,17,19,20,35,36,38,44–51 Gilboa et al.52 studied handwriting, a skill that is often reported to be a problem for children with NF1, and showed that children with NF1 scored significantly lower on all measures of the Six-Trait Writing Model test (ideas, organization, word choice, sentence fluency, conventions, and total score) than comparison participants.
However, not all studies have found impairment on motor tasks: Bawden et al.42 and Billingsley et al.23 did not find impaired performance on the Grooved Pegboard; Eldridge et al.22 did not find an impairment in the Beery Visual–Motor Integration test; Krab et al.48 failed to show a significant difference on the Prism Adaptation test; and Moore et al.25 found no significant differences on any of the measures they used, although children with NF1 had lower scores than comparison participants. Also, although the handwriting of children with NF1 in the study by Gilboa et al.52 was less legible, less well-organized, and included more letter corrections and illegible letters than the handwriting of comparison children, these differences did not reach statistical significance.
There are some possible explanations for the variability in the results. Motor function, executive, and visuospatial skills are closely connected, and, in the case of handwriting, linguistic skills would also play a part; these are not isolated skills that can be measured discretely. Instead, it would be necessary to measure all these skills within one study and to investigate their interrelationships to differentiate their contribution to overall performance. Furthermore, children with NF1 have been documented to have decreased mineral density, bone strength, and muscle mass,53 and may have reduced muscle force.54 It is possible that these deficits affect children’s performance in motor tasks as well as their motor functioning in everyday life. More research into these connections is needed.
Intelligence and thinking
Intelligence as a construct has been defined a number of times in different contexts. An editorial statement of 52 researchers in 1994 defined it as: ‘A very general mental capability that, among other things, involves the ability to reason, plan, solve problems, think abstractly, comprehend complex ideas, learn quickly and learn from experience. It is not merely book learning, a narrow academic skill, or test-taking smarts. Rather, it reflects a broader and deeper capability for comprehending our surroundings—“catching on,”“making sense” of things, or “figuring out” what to do.’55
Relying solely on measurements of intelligence or IQ would no doubt lead to oversimplification of children’s abilities. The skills and experiences that influence an individual’s performance on intelligence tests are too diverse to comprehensively summarize in one or a few composite scores. However, from a clinical point of view it is useful to determine an individual’s general level of functioning and the pattern of strengths or weaknesses. These can then be investigated further if necessary. Furthermore, intelligence tests are composed of a number of subtests that measure various thinking skills, which are one aspect of the cognitive skills of the individual.
Cognitive processes measured by IQ tests: thinking skills
Lezak et al.’s5 conceptualization of cognitive functions includes thinking (involving different ways to mentally organize or manipulate information). Many of the subtests of standard IQ tests involve reasoning and concept formation, and therefore shed light on the skills of children with NF1 in these specific areas of cognition.
Verbal concept formation can be measured with the similarities subtest of the WISC. Hyman et al.,3 Ferner et al.,56 Hofman et al.,16 and Mazzocco et al.19 all found that children with NF1 performed significantly worse on this subtest than sibling comparisons, while Krab et al.17 did not find marked lowering, although they did not report statistical comparisons for their whole group against a normative sample. Furthermore, 45% of children with NF1 scored more than 1SD below their sibling on this task.
Reasoning skills are measured by several of the WISC subtests: comprehension (verbal reasoning), picture completion, and picture arrangement (visuoperceptual reasoning) and arithmetic (mathematical reasoning). Picture arrangement was the only subtest on which Hyman et al.3 did not find children with NF1 to be impaired, while Mazzocco et al.19 did not find significant differences between children with NF1 and siblings on any of these four subtests. Their participant numbers were small, however so this may in part explain the lack of significant differences. Hofman et al.16 report results only for the arithmetic subtest, and showed that children with NF1 were not impaired. Ferner et al.56 do not report results for any of these four subtests. Finally, in the study by Krab et al.,17 children with NF1 did not seem to particularly struggle with picture completion or picture arrangement, but performed less well on arithmetic and comprehension.
IQ profile of children with NF1
The full-scale IQ of the majority of children with NF1 is in the normal range, although somewhat lower than children in normative samples or comparison groups (usually around 90, instead of the normative mean of 100).3,17–20,22,23,27,33,36,40,42,44,56–61 Only around 6% to 7% of children with NF1 have an IQ lower than 70,3,36,56 compared with only 2% of a normative population. The mean verbal and performance IQ of the participants with NF1 is usually significantly lower than that of comparison participants. Despite the apparent robustness of these findings, some studies have failed to demonstrate significant differences in IQ between children with NF1 and comparison children or norms.15,16,24,25,31 Apart from the studies by Moore and colleagues,24,25 all studies that failed to find an IQ difference between children with NF1 and comparison participants included only small numbers of participants, and the results may therefore be due to lack of power or, alternatively, biased sampling. Although several studies have looked at discrepancies between verbal and performance IQ, a consensus on the matter has not been reached, as different studies have found either no discrepancy or discrepancies to different extents.3,36,60 Indeed, North et al.8 suggested that the discrepancies are of ‘questionable significance’.
There is disagreement with regards to the exact IQ profile of children with NF1. Some studies have found that children with NF1 perform significantly more poorly on almost all subtests of the WISC3,16,56 with the exception of arithmetic16 and picture arrangement.3 The results of Mazzocco et al.19 were somewhat less conclusive; they found significant differences between all the verbal comprehension subtests except comprehension, while on only one perceptual/organization subtest, block design, were significant differences between children with NF1 and their siblings detected. In addition, digit span was significantly different between the two groups. Krab et al.17 do not report the significance of the comparisons between the NF1 group and a normative sample, but the children’s performance is close to the expected normative mean (mean standard scores over 9, mean 10) on similarities, picture completion, picture arrangement, and coding. Their results are therefore not entirely dissimilar from the findings of Hofman et al.16
Memory is the ability to ‘retain information and utilize it for adaptive purposes’.62 Memory functions are often divided to declarative memory (explicit memory, semantic, and episodic memories) and implicit memory (procedural memory and implicit item-specific memory). A distinction between verbal and non-verbal memory is also drawn.
In general, studies focusing on memory have given mixed results. Bawden et al.42 used the Warrington Recognition Memory test and the Verbal and Nonverbal Selective Reminding Tests, and found that children with NF1 were impaired on the recall measure of Selective Reminding tests and on the long-term storage score of Verbal Selective Reminding. Billingsley et al.23 found deficits in verbal but not spatial memory, while Moore et al.24 did not find significant differences on the memory measures (although they did not look at the verbal/non-verbal subtests separately). Hyman et al.3,36 found no significant differences between the NF1 and sibling groups on the verbal or visual memory tests they used (the Continuous Visual Memory Test and the California Verbal Learning Test for Children). Likewise, Krab et al.17 found that children with NF1 were not markedly impaired on the Rey Auditory Verbal Learning Test and Descheemaeker et al.35 found no significant impairment on the 15 words of Rey (immediate or delayed recall). Ullrich et al.40 found no differences between NF1 and comparison groups on the California Verbal Learning Test for Children. A slightly more consistent finding concerns the delayed recall condition of the Rey Osterrieth Complex Figure test, which has been used to measure memory; children with NF1 tend to perform poorly on delayed recall.16,17,35,42
Ullrich et al.40 studied visuospatial learning and memory by devising a computerized, virtual reality task equivalent to the Morris Water Maze task, performance on which is known to be impaired in the NF1 mouse model. With a visible target, there were no differences between the children with NF1 and comparison participants in speed and direction. However, when the target was hidden from view, children with NF1 spent significantly less time than sibling comparisons searching the correct quadrant. It is worth noting that children who performed poorly on the working memory tasks were especially impaired on the virtual maze task, which may indicate that this task too may involve an executive function component. Overall, however, Ullrich et al.’s40 study provides an example of a task that is ecologically valid (i.e. it approximates a real-life setting). However, the sample size was very small and therefore the results should be replicated with a larger sample.
Executive function and attention
Lezak et al.5 define executive functions as capacities that enable a person to successfully engage in ‘independent, purposive, self-serving behaviour’, and Gioia et al.63 further add that executive functions guide and manage cognitive and emotional functions, especially in the context of new situations or problems. Executive functions include working memory, planning, organization, inhibitory processes, categorization, flexibility, rule deduction, and divided and sustained attention.64–66
Because executive functions encompass several functions, there is a great deal of variety in the focus and measures that different studies have adopted to investigate whether executive functions are affected by NF1. Some have concentrated on working memory; Ferner et al.56 used a working memory task (the Sternberg Memory Test) and found that the NF1 participants made more errors than comparison children, although the reaction times of the two groups did not significantly differ. Children with NF1 performed more poorly than their siblings on the CANTAB Spatial Working Memory task40 and on the working memory task from the Amsterdam Neuropsychological Tasks battery.45
Many of the studies on children with NF1 have investigated the planning aspects of executive functions; overall their findings suggest that this aspect of executive function is impaired in children with NF1. Hyman et al.3 and Pride et al.27 found that children with NF1 performed significantly more poorly than siblings on tasks of executive functioning (including the Tower of London and the Children’s Category Test). Over 70% of the children were more than 1SD below the norm on the Tower of London test. Roy et al.58 studied the planning abilities of children with NF1 and obtained similar results, demonstrating that the NF1 group performed more poorly than comparison participants on all measures of planning skills.
Inhibitory control is another facet of executive function affected in children with NF1.45,47 Ferner et al. found that children with NF1 were impaired on the Stroop task; in around 20% of the NF1 participants, mean performance on this task was 2SD below the mean for comparison participants. Moreover, both Huijbregts et al.45 and Rowbotham et al.47 showed that cognitive control is important (defined as an equivalent of executive function, but also encompassing the level of complexity required to perform a task). Children with NF1 performed particularly poorly on tasks of working memory and inhibitory control when the demand for cognitive control increased; the effect on motor tasks was less clear. Rowbotham et al.47 found that the performance of children with NF1 deteriorated more substantially than that of comparison participants when processing load increased. Children with NF1 also made more errors when flexibility was required. In contrast to the findings discussed above, Krab et al.17 and Descheemaeker et al.35 found no marked deficits in executive functioning (Wisconsin Card Sorting Task, Stroop task), although children with NF1 performed slightly below the norm.
Traditionally, executive functions have been measured directly. However, a questionnaire instrument, the Behaviour Rating Inventory of Executive Function, can also be used and is argued to be more ecologically valid than specific psychometric tests. Pride et al.27 found that children with NF1 scored poorly on the shift, emotional control, and monitor subscales, as well as the Global Executive Composite and Behavioural Regulation Index. Payne et al.26 reported significantly lower scores in the NF1 group than in the comparison group on all the scores of the Behaviour Rating Inventory of Executive Function. Notably, these differences remained significant even after controlling for verbal IQ.
According to Lezak et al.,5 attentional functions are processes concerning the way in which the individual responds to stimuli and they do not have a unique behavioural end product as such. They have limited resources,67 are organized in a hierarchical manner, and take place sequentially.68
As with executive functions, attention problems in children with NF1 have been addressed by using neuropsychological tests to directly measure the deficits that children with NF1 have and by investigating how attention problems affect children’s behaviour and functioning in their daily life (using various questionnaire measures). A number of studies3,17,19,28,35,56 (although not Dilts et al.15) have shown that children with NF1 are impaired on attention tasks. Hyman et al.3 found that children with NF1 were impaired only on the sustained and switching attention tasks of the Tea-Ch battery, not on all subtests. Pride et al.27 and Watt et al.28 replicated this result, although Payne et al.26 found that children were impaired on all of the screening measures on the Tea-Ch battery. Gilboa et al.69 used a virtual reality paradigm in which they embedded a version of the continuous performance task into a computer-based classroom simulation with typical distracters (auditory and visual). They found that children with NF1 made more commission errors and had fewer total correct hits than comparison children. The children’s performance also correlated negatively with their inattention/cognitive problems score on the Conners’ Parent Rating Scale.
The majority of studies that have measured attention symptomatology with functional questionnaires have found that a substantial number of children with NF1 scored highly enough on these questionnaires to meet the researchers’ criteria for ADHD.16,21,22,33,40,58,70,71 Overall, approximately one-third to one-half of children with NF1 fulfilled the criteria set for ADHD (or poor attention) according to the research criteria employed by different studies. Koth et al.70 showed that ADHD was much more common in children with NF1 than in their siblings (or parents). However, it is important to note that scoring high on a questionnaire is not enough to warrant a diagnosis (especially as the questionnaires that are utilized vary widely in quality), and additional observational, interview, and psychometric measures would always be employed before arriving at a formal diagnosis. This procedure has been most closely followed in a research context by Hyman et al.,3 who used a combination of parents’ and teachers’ Conners’ questionnaire scores, as well as children’s performance on the psychometric tasks and observations during the assessment. In addition, Mautner et al.72 used a diagnostic interview to investigate ADHD in their sample of 93 children, of whom 49.5% were diagnosed with the disorder.
Emotionality, behaviour problems, and social competence
Lezak et al.5 point out that emotionality plays an important part in determining the behaviour of an individual. Emotionality is the ‘observable behavioural and physiological component of emotion’73 and emotion is a system that allows the individual to respond to events rapidly and appropriately.74 Emotion is associated with mood, temperament, disposition, motivation, and personality.75 Emotions guide our thoughts and actions and regulate our behaviour, and therefore help us to appropriately adapt to our environment.76 Emotion regulation involves processes such as cognitive appraisal (reading and understanding social cues, perception of affect, predicting one’s own and others’ actions); expression of emotions (communicating feelings to others during social interaction); socialization of emotions (reinforcement to express certain emotional displays in certain contexts); and modulation of emotions and mood states (in response to internal states, environmental demands and the social context77). Various cognitive and executive functions are involved in several aspects of emotion regulation, which means that deficits in cognition and/or executive function can potentially affect an individual’s abilities to regulate his/her emotions. Other factors that can potentially affect emotion regulation78 are child temperament, emotional development, language delay, general developmental delay, overactivity, parental depression, environmental stress, parental management techniques, and parental history.
Emotion and behaviour problems
Behaviour problems are an issue that many parents of children with NF1 bring up in clinic. Research investigating the type and prevalence of behaviour problems in children with NF1 has mainly used different questionnaires. The findings from the CBCL across studies suggest that, when rated by parents, children with NF1 have a significantly higher score on the total problems scale of the CBCL than comparison children or the CBCL norms.15,79–81 Descheemaeker et al.35 found that, although the mean scores for the group of children with NF1 in their study were in the normal range overall, 35% of the children were rated in the ‘at risk’ or ‘clinical’ ranges by parents whereas only 2% of typically developing children would be expected to fall into the ‘clinical’ range. The studies also showed that children with NF1 scored significantly higher than the norms on internalizing, externalizing, and thought problems.79,81 The children in the study by Dilts et al.15 scored significantly more highly than comparison children on anxious/depressed, internalizing, and attention problems. The teacher ratings for the CBCL differ from the parent ratings in that teachers rated children with NF1 significantly higher than the norms on total problem behaviours and internalizing problems; however, there were no differences when comparisons were made between siblings and children with NF1.79 Dilts et al.,15 on the other hand, found that teachers rated children with NF1 significantly higher on anxious/depressed problems than they did comparison children. Huijbregts and de Sonneville82 and Johnson et al.,4 using the Strengths and Difficulties Questionnaire, found that children with NF1 had significantly more conduct and emotional problems than children in the comparison group, and a significantly higher proportion of children with NF1 than norms scored in the ‘clinical’ range on all the subscales of the CBCL apart from prosocial behaviour.4 In contrast, Noll et al.80 found that there were no differences between children with NF1 and comparison children on ratings of depression and loneliness.
Social competence: social problems and social skills
Social competence is not discussed by Lezak et al.5 as one of the three fundamental systems contributing to behaviour. We have included it in this review because social competence is an important aspect contributing to an individual’s overall function. In defining social competence, we follow Rose-Krasnor’s83 model, which emphasizes the importance of effectiveness in social interactions as key in social competence. She sees social competence as situation specific and influenced by the systems affecting behaviour: cognitive and executive functions, emotionality, and attention. A child’s developing skills in the areas of motor, memory, linguistic, executive, and other functions will affect his or her competence. Thus, developmental problems in any of these areas are likely to have repercussions on the child’s social competence.
Studies investigating social competence in children with NF1 have usually focused on measuring either social problems (undesirable behaviours such as clinginess, acting young) or social skills (socially acceptable learnt behaviours, e.g. giving a compliment84), or abilities such as perspective taking, empathy, and social problem-solving.85 Both social skills and social problems contribute to a child’s social competence. The majority of data concerning the social skills and social problems of children with NF1 have come from questionnaires: data from the CBCL, the Strengths and Difficulties Questionnaire, the Social Skills Rating System (SRSS), and the Social Responsiveness Scale all suggest that social problems are common in children with NF1.15,79–82 Huijbregts and de Sonneville82 also showed that children with NF1 scored significantly more highly than comparison children on autistic traits (Social Responsiveness Scale); group differences were largest for this variable. Finally, Noll et al.80 reported that teachers perceived children with NF1 as more prosocial than comparison classmates (using the Revised Class Play instrument). In contrast, Barton and North79 found no significant differences on social skills (measured by the SRSS) between children with NF1 and their siblings on any of the variables they looked at.
Studies have also measured social competence more directly. Barton and North79 studied the association of social skills (from the SRSS ratings) and social outcomes, based on the social problem and the social competence scores from the CBCL (e.g. taking part in sports, hobbies, clubs, and teams). They found that children with NF1 scored significantly lower than their siblings on all social competence scales, thus replicating the results of Dilts et al.15 and Barton and North79 also found that parents’ and teachers’ ratings were correlated with the ratings of social competence and the behavioural correlates of social skills: better social skills were associated with significantly fewer social problems and better social competence. Noll et al.80 investigated social behaviour and peer acceptance in children with NF1 through peer and teacher ratings. Results from the Revised Class Play instrument (peers/teachers asked to assign roles to different children within the classroom) suggested that children with NF1 were perceived by both teachers and peers as more sensitive and isolated, less well liked, displayed less leadership behaviour, were selected less often as best friends, and had fewer reciprocal friendships than comparison children. Interestingly, children with NF1 did not have lower self-ratings on the Revised Class Play than children without NF1. Thus, overall, evidence from various sources seems to agree that children with NF1 have poorer social competence than comparison children.
The reasons for the poorer social competence of children with NF1 warrant more research. Barton and North79 did not find IQ or academic achievement to be related to social competence, whereas Huijbregts et al.86 found that there was a significant difference between the social skills of children with NF1 and comparison children, which disappeared once cognitive processing abilities had been accounted for. However, it is important to note that Huijbregts et al.86 used a composite measure of cognitive processing, whereas Barton and North79 used IQ and academic achievement as measures. Another potential factor is the severity of physical NF1 symptoms. These, however, do not appear to consistently affect the social competence of children with NF1 as Barton and North79 found that, in children’s self-ratings, children with moderate/severe NF1 had poorer social skills, while neither Huijbregts and de Sonneville82 nor Johnson et al.81 demonstrated this association.
More recently, Huijbregts et al.86 investigated whether children with NF1 have deficits in social information processing (e.g. recognition of faces, facial expressions and emotions, or top-down processes involving stored knowledge for adjusting one’s behaviour).86 Huijbregts et al.86 investigated whether children with NF1 are impaired in their ability to process social information using tasks from the Amsterdam Neuropsychological Tasks battery that tapped face and emotion recognition. They found that, in general, children and young people with NF1 are slower and more variable in their performance when asked to recognize faces in front and in profile, to identify fear, and to match emotions. This would suggest that children with NF1 may have difficulties when bottom-up encoding of social information is required. Importantly, however, using cognitive control (tasks of inhibitory control and working memory) as a covariate in the analyses reduced the significance of group differences on processing of social stimuli; differences remained significant in fear recognition, accuracy of emotion matching, accuracy in recognizing frontal faces, and stability in recognizing faces in profile.86 The authors consequently argued that cognitive control is involved in the processing of social information.