Introduction
Neurofibromatosis type 1 (NF1) is a genetic disease affecting around 1/3500 births. It is caused by a mutation of the 17q11.2 gene coding for neurofibromin, which normally plays an onco-suppressive role. Owing to its autosomal dominant nature, the disease is transmitted by one of the two parents in 50% of cases (familial form), while in the other half of patients, the mutation occurs spontaneously during pregnancy (sporadic form). Patients can exhibit a range of clinical signs (dermatological, ophthalmological, and/or orthopedic), which also constitute the diagnostic criteria established by the NIH (Gutmann et al., Reference Gutmann, Aylsworth, Carey, Korf, Marks, Pyeritz, Rubenstein and Viskochil1997; revised by Legius et al., Reference Legius, Messiaen, Wolkenstein, Pancza, Avery, Berman, Blakeley, Babovic-Vuksanovic, Cunha, Ferner, Fisher, Friedman, Gutmann, Kehrer-Sawatzki, Korf, Mautner, Peltonen, Rauen, Riccardi, Schorry, Stemmer-Rachamimov, Stevenson, Tadini, Ullrich, Viskochil, Wimmer, Yohay, Gomes, Jordan, Mautner, Merker, Smith, Stevenson, Anten, Aylsworth, Baralle, Barbarot, Barker, Ben-Shachar, Bergner, Bessis, Blanco, Cassiman, Ciavarelli, Clementi, Frébourg, Giovannini, Halliday, Hammond, Hanemann, Hanson, Heiberg, Joly, Kalamarides, Karajannis, Kroshinsky, Larralde, Lázaro, Le, Link, Listernick, MacCollin, Mallucci, Moertel, Mueller, Ngeow, Oostenbrink, Packer, Papi, Parry, Peltonen, Pichard, Poppe, Rezende, Rodrigues, Rosser, Ruggieri, Serra, Steinke-Lange, Stivaros, Taylor, Toelen, Tonsgard, Trevisson, Upadhyaya, Varan, Wilson, Wu, Zadeh, Huson, Evans and Plotkin2021). The severity and visibility of NF1 can be quantified for each patient. In terms of severity, the Riccardi scale classifies patients according to the extent of skin involvement and the presence of disabling complications (Riccardi, Reference Riccardi1982). The Ablon visibility index is employed to assess the esthetic consequences of NF1 (Ablon, Reference Ablon1996).
Among the various complications associated with the disease, neuropsychological disorders are frequently reported by patients. Intellectual efficiency is relatively preserved, with only a downward shift toward the low average. Conversely, executive and attentional impairments are frequent and may negatively influence the expression of other cognitive abilities such as visuospatial abilities, language, and praxis (Descheemaeker et al., Reference Descheemaeker, Plasschaert, Frijns and Legius2013; Ferner et al., Reference Ferner, Hughes and Weinman1996; Pavol et al., Reference Pavol, Hiscock, Massman, Bartlett Moore, Foorman and Meyers2006; Zöller et al., Reference Zöller, Rembeck and Bäckman1997). There is considerable heterogeneity between patients with NF1 for both physical and neuropsychological symptoms.
Patients frequently report social difficulties in daily life, such as loneliness or low peer acceptance (for a review, see Chisholm et al., Reference Chisholm, Anderson, Pride, Malarbi, North and Payne2018). A number of biopsychosocial theoretical models support the idea that adaptation and social functioning are closely linked to the interaction between factors specific to the individual and his environment, with a number of cognitive functions, including executive functioning and social cognition (Beauchamp & Anderson, Reference Beauchamp and Anderson2010; Yeates et al., Reference Yeates, Bigler, Dennis, Gerhardt, Rubin, Stancin, Taylor and Vannatta2007). Furthermore, some authors have therefore assumed that social adaptation is negatively influenced by social cognition disorders in NF1 (Huijbregts et al., Reference Huijbregts, Jahja, De Sonneville, De Breij and Swaab-Barneveld2010). Social cognition refers to a set of cognitive and affective abilities that allow for the perception and processing of social information, with a view to producing an appropriate response to an interaction situation (e.g., Adolphs, Reference Adolphs2009). It includes so-called low-level functions such as emotion recognition, as well as more complex processes such as theory of mind (ToM), moral reasoning, and social information processing. Emotion recognition corresponds to the processing of information needed to identify the emotions felt by others (Ekman & Friesen, Reference Ekman and Friesen1976). ToM enables individuals to understand and/or predict the mental states and behaviors of others (Dvash & Shamay-Tsoory, Reference Dvash and Shamay-Tsoory2014). Moral reasoning corresponds to the way in which individuals consider the world around them and judge the actions of others as right or wrong (Ehrlé et al., Reference Ehrlé, Henry, Pesa and Bakchine2011; Greene & Haidt, Reference Greene and Haidt2002). Social judgment and social information processing refer to the ability to determine whether a behavior or situation complies with social norms and/or to attribute social meaning to external stimuli (Beauchamp & Anderson, Reference Beauchamp and Anderson2010; Ehrlé et al., Reference Ehrlé, Henry, Pesa and Bakchine2011).
To date, only two studies have explored social cognition processes in adults with NF1 (Pride et al., Reference Pride, Crawford, Payne and North2013, Pride et al., Reference Pride, Korgaonkar, Barton, Payne, Vucic and North2014). These yielded contradictory data pointing to emotion recognition difficulties. One study found specific weakness of anger recognition (Pride et al., Reference Pride, Korgaonkar, Barton, Payne, Vucic and North2014), while the other reported that participants had greater difficulty recognizing positive emotions (Pride et al., Reference Pride, Crawford, Payne and North2013). Pride et al. (Reference Pride, Korgaonkar, Barton, Payne, Vucic and North2014) also examined ToM in adults with NF1, highlighting weaknesses in mentalizing. Moral reasoning and social information processing have not yet been studied in adults with NF1, but a study of children with NF1 showed significant differences in performance on tasks assessing these processes compared to a control group, raising questions about these aspects in adulthood (Remaud et al., Reference Remaud, Besnard, Barbarot and Roy2024).
Some researchers have postulated that specific factors arising from the clinical context influence sociocognitive behaviors in a range of pathologies (e.g., Beauchamp & Anderson, Reference Beauchamp and Anderson2010). However, the links between NF1-specific features (mode of inheritance, severity, and visibility) and social cognition have rarely been explored. The few studies to have investigated the links between mode of inheritance and social abilities or social cognition in NF1 were conducted among children and failed to find any association (Chisholm et al., Reference Chisholm, Anderson, Pride, Malarbi, North and Payne2018; Remaud et al., Reference Remaud, Besnard, Barbarot and Roy2023). To our knowledge, these issues have not yet been explored in adults with NF1 and therefore require further investigation.
Similarly, the relationship between severity and visibility of the disease and social cognitive functioning has not been studied in adults with NF1. However, in terms of severity, several studies that have looked at the relationship between patients’ severity and their social functioning have found no relationship between the measures (for a review, see Chisholm et al., Reference Chisholm, Anderson, Pride, Malarbi, North and Payne2018). It should be noted that the severity of the disease in these studies was measured using the Riccardi scale, which refers only to the medical severity and physical health of the patient. Other studies, such as that by Noll et al. (Reference Noll, Reiter-Purtill, Moore, Schorry, Lovell, Vannatta and Gerhardt2007), have included neurological and neuropsychological complications in addition to the Riccardi scale to measure the severity of NF1, and have highlighted weaker social skills in the most severely affected children.
With regard to visibility, an indirect association has been demonstrated between the level of visibility of NF1 adults and their quality of life (Sanagoo et al., Reference Sanagoo, Jouybari, Koohi and Sayehmiri2019; Vranceanu et al., Reference Vranceanu, Merker, Park and Plotkin2013), including their social and emotional quality of life (Doser et al., Reference Doser, Andersen, Kenborg, Dalton, Jepsen, Krøyer, Østergaard, Hove, Sørensen, Johansen, Mulvihill, Winther and Bidstrup2020). Moreover, one study found that NF1 patients with more visible symptoms reported that their skin condition had a greater impact on their emotions and social functioning (Page et al., Reference Page, Korf, Page, Leplege and Wolkenstein2004). Furthermore, visible manifestations of the disease, such as cutaneous neurofibromas, are postulated to exert a pivotal influence on patients’ self-perception (Hummelvoll & Antonsen, Reference Hummelvoll and Antonsen2013; Smith et al., Reference Smith, Wang, Plotkin and Park2013). A substantial body of research has delineated the intimate interconnections between self-perception or self-awareness and perceptions of others. The findings of these studies indicate that the representations of the mental states of others are based on an analysis of intentions, knowledge, and beliefs about oneself (Stuss et al., Reference Stuss, Gallup and Alexander2001). Furthermore, in his neuroanatomical model of the frontal lobes, Stuss (Reference Stuss2011) delineated multiple levels, the fourth of which corresponded to “self-awareness.” This level enables self-perception and implies the metacognitive ability to utilize one’s own experience to comprehend the mental states of others, which aligns with theory of mind. Accordingly, their perspective posits that “self-awareness” component encompasses both self-awareness and awareness of one’s own social position, thereby facilitating accurate interpretation of the world. Thus, we can speculate that increased visibility of the disease would have repercussions for patients’ self-perception and self-awareness, which in turn would have repercussions for their theory of mind abilities, and more broadly on social cognition.
The main aim of the present study was to explore the social cognition abilities (emotion, ToM, moral judgment, and social information processing) of adults with NF1. A secondary objective was to investigate the impact of disease characteristics (mode of inheritance, severity, and visibility) on these abilities. Preliminary data (from publications in adults and/or children with NF1) suggested that adults with NF1 score lower on emotion processing, ToM, moral judgment, and social information processing than healthy individuals. In view of the limited and/or indirect data (from the pediatric population or on quality of life), we expected patients with higher visibility to exhibit greater social cognition difficulties on all measures, but predicted that mode of inheritance and severity would have no impact.
Method
Participants
Patients were recruited from a center specializing in the management of neurofibromatosis within a university hospital. All patients seen at the center underwent a systematic neuropsychological assessment at a learning disabilities reference center as part of their follow-up. Data were then processed retrospectively. There were six inclusion criteria: (1) fulfilment of diagnostic criteria of the National Institute of Health Consensus Conference (Gutmann et al., Reference Gutmann, Aylsworth, Carey, Korf, Marks, Pyeritz, Rubenstein and Viskochil1997, revised in 2021 Legius et al., Reference Legius, Messiaen, Wolkenstein, Pancza, Avery, Berman, Blakeley, Babovic-Vuksanovic, Cunha, Ferner, Fisher, Friedman, Gutmann, Kehrer-Sawatzki, Korf, Mautner, Peltonen, Rauen, Riccardi, Schorry, Stemmer-Rachamimov, Stevenson, Tadini, Ullrich, Viskochil, Wimmer, Yohay, Gomes, Jordan, Mautner, Merker, Smith, Stevenson, Anten, Aylsworth, Baralle, Barbarot, Barker, Ben-Shachar, Bergner, Bessis, Blanco, Cassiman, Ciavarelli, Clementi, Frébourg, Giovannini, Halliday, Hammond, Hanemann, Hanson, Heiberg, Joly, Kalamarides, Karajannis, Kroshinsky, Larralde, Lázaro, Le, Link, Listernick, MacCollin, Mallucci, Moertel, Mueller, Ngeow, Oostenbrink, Packer, Papi, Parry, Peltonen, Pichard, Poppe, Rezende, Rodrigues, Rosser, Ruggieri, Serra, Steinke-Lange, Stivaros, Taylor, Toelen, Tonsgard, Trevisson, Upadhyaya, Varan, Wilson, Wu, Zadeh, Huson, Evans and Plotkin2021); (2) aged between 18 and 59 years 11 months; (3) no current or history of neurological and/or psychiatric disorders, epilepsy, or brain tumor (it should be noted that patients with benign optic pathway gliomas were not excluded from the study because of the frequency of the symptom and its asymptomatic nature); (4) no sensory disorders incompatible with a neuropsychological assessment; (5) sufficient command of the French language to perform the tests; and (6) no neuropsychological assessment within the previous 6 months. Of the 53 adults initially invited to take part, 19 were excluded because of neurological and/or psychiatric comorbidities (i.e., depression, epilepsy, brain tumors, and/or Gilles de la Tourette syndrome), 2 had sensory disorders incompatible with the tests, and 12 refused to take part. The final sample therefore comprised 20 adults with NF1.
A control group was formed of healthy adults matched with the patient group for age, gender, and education level. The latter corresponded to the number of completed years of study, from first grade to the highest diploma (e.g., high-school diploma corresponded to 12 years of study). Inclusion criteria were the same as for patients, with the exception of the NF1 diagnosis. Of the 23 adults we interviewed, three were excluded, to ensure homogeneity of sociodemographic variables across the two groups. The final control group therefore comprised 20 adults.
Material
We administered the French Social Cognition Battery (FSCB; Ehrlé et al., Reference Ehrlé, Henry, Pesa and Bakchine2011), which has an emotional component and a social inference component. The initial data of studies with different clinical groups, such as multiple sclerosis (Ehrlé et al., Reference Ehrlé, Hody, Lecrique, Gury and Bakchine2020; Henry et al., Reference Henry, Tourbah, Chaunu, Rumbach, Montreuil and Bakchine2011, Reference Henry, Bakchine, Maarouf, Chaunu, Rumbach, Magnin, Tourbah and Montreuil2015, Reference Henry, Tourbah, Chaunu, Bakchine and Montreuil2017) and frontal dementia (Ehrlé et al., Reference Ehrlé, Henry, Pesa and Bakchine2011, Reference Ehrlé, Cavalier, Toulon, Delaby, Menu and Bakchine2015), indicate a good sensitivity of the battery for the evaluation of social cognition in adults. For the purpose of the present study, we administered three emotional tasks (Emotion Identification, Expressive Intensity Judgment, and Emotion Discrimination) and three inferential tasks (Theory of Mind, Moral and Conventional Judgments, and Social Situations).
In the Emotion Identification task, pictures of a person’s face (Ekman & Friesen, Reference Ekman and Friesen1976) expressing the six basic emotions (joy, surprise, anger, fear, disgust, and sadness) are shown one by one. Participants must name the emotion expressed in each picture as quickly as possible. There are 10 trials per emotion, making a total of 60 pictures. The total success score (0–60) and scores per emotion (0–10) are recorded, together with participants’ response times.
The Expressive Intensity Judgment task assesses the arousal level of the emotion being expressed. It consists of 30 trials in which pictures of a young woman expressing the six basic emotions are displayed one by one, with each picture expressing an emotion of a different intensity. In order to construct the items, photographs of Ekman’s F7 young woman (Ekman & Friesen, Reference Ekman and Friesen1976) expressing the six basic emotions were morphed with the same person’s neutral expression. A total of 20 levels of intensity were generated for each continuum. In this task, five levels of expressive intensity (1, 2, 3, 4, and 5) were retained, corresponding to levels 3, 7, 11, 15, and 19 of the morphed version. Participants are asked to estimate the intensity of the emotion (without having to recognize it) on a scale ranging from 1 (low intensity) to 5 (high intensity). The average intensity estimated per emotion and mean response times are recorded.
The Emotion Discrimination task involves the perceptual dimension to a greater extent. Pictures of the young woman expressing primary emotions are displayed in pairs, and participants have to indicate whether the two pictures represent the same emotion or two different ones (without identifying them). Of the 54 trials, 24 involve identical pairs, and 30 involve nonidentical ones. Identical pairs were made up of stimuli evoking the same emotion but with varying expressive intensities. Four items were proposed for each emotion. In the case of pairs comprising different emotions, each emotion was presented in conjunction with each of the other five primary emotions. All the emotions in the different pairs exhibited identical expressive intensities. The total success score and the time taken to complete the task are recorded for each participant.
The Theory of Mind task assesses the ability to attribute first- and second-order beliefs, as well as to detect missteps (i.e., identification of a social blunder committed by one character towards another). Seven different short stories (text and pictures) are presented to participants in a random order. Two stories concerned attribution of first-order beliefs, two others second-order beliefs, and two still others missteps. The final story is a physical inference control item. Participants are permitted to review each story as many times as they wish. The examiner then asks them questions about the story and the characters’ mental state. A control question is asked for each story, relating to the non-social content of the story, to check that participants have understood it correctly. Some questions involving a binary response are scored 0 or 1 point. For other responses, an imprecise or incomplete response that is not erroneous may be given a score of 0.5 points. Subsequently, a score for first-order beliefs, second-order beliefs, and missteps is calculated, corresponding to the sum of the scores obtained in the corresponding stories.
The Moral and Conventional Judgments task is based on tests created by Blair and Cipolotti (Reference Blair and Cipolotti2000) and Turiel et al. (Reference Turiel, Killen, Helwig, Kagan and Lamb1987). Sentences describing acts carried out by characters are provided orally (read by the examiner) and in writing. A total of 15 items are randomly presented to the participant, comprising five items presenting non transgressive acts, five items of transgressing acts according to conventional norms, and five items that transgress moral norms (e.g. “a woman laughs at a person with disabilities”). Participants are then asked to judge the permissive nature of the act (“Is it right or wrong for a woman to laugh at a person with disabilities?”). If they answer “wrong,” they are then asked to rate the seriousness of the act (“How wrong is it for a woman to laugh at a person with disabilities?”) on a scale ranging from 1 (Not very serious) to 5 (Very serious), and to give a justification/explanation (“Why is it wrong for a woman to make fun of a person with disabilities?”; data not shown). The last two questions assess the authoritative nature of each moral or conventional rule in terms of its generalization (absence of a rule against or prohibiting the act : “In a country where there was no law prohibiting it, would it be right or wrong for a woman to laugh at a person with disabilities?”) or dependence (act authorized by a legitimate authority : “A recognized scientist says that anyone is allowed to laugh at a cripple. In these circumstances, is it right or wrong for a woman to laugh at a person with disabilities?”). For the purpose of this study, we retained scores for initial success (good/bad), severity, and generalization and dependence for the moral and conventional transgression conditions (each score ranging from 0 to 5).
Adapted from tests developed by Blair and Cipolotti (Reference Blair and Cipolotti2000) and Dewey (Reference Dewey and Frith1991), the Social Situations task is designed to assess the ability to judge social norms. Eight short stories are presented to participants in the form of texts, which are read out at the same time by the examiner. In these narratives, two sentence are presented in italics at two distinct points in the narrative (Figure 1). The participant is asked to judge whether the behavior/action described in italics seems appropriate or inappropriate in the context of the narrative. Two judgments must be made on two behaviors for each story. These texts describe a totally of 8 appropriate (i.e., behavior that can be considered to be almost normal in the context of the proposed situation) versus 8 inappropriate behaviors (i.e., rather strange or eccentric behavior, which may induce fear or annoyance) in social situations (Ehrlé et al., Reference Ehrlé, Henry, Pesa and Bakchine2011). The total success score (0–16), and total number of correct answers given per condition (adapted vs. non-adapted; 0–8) was used as the variable, corresponding to the behavior for which the participant had correctly identified the adapted or non-adapted nature of the action.
Figure 1. Illustration of the ‘Social situations’ task adapted from Ehrlé et al. (Reference Ehrlé, Henry, Pesa and Bakchine2011).
We also assessed intellectual functioning and attentional abilities, using the Wechsler Adult Intelligence ScaleFourth Edition (WAIS-IV; in its entirety for patients, and including only abbreviated measures of the Matrices and Vocabulary subtests for controls), and the Rating Scale of Attentional Behavior (RSAB) in self- and informant-report questionnaires, in order to control for disorders that might influence participants’ performance on the social cognition tests.
With regard to disease characteristics, severity levels were measured using the Riccardi scale (Riccardi, Reference Riccardi1982), with scores ranging from 1 (minimal) to 4 (severe). Visibility was assessed using the Ablon scale (Ablon, Reference Ablon1996), with scores ranging from 1 (poorly visible) to 3 (highly visible; see Table 1 for a description).
Table 1. Riccardi and ablon index (PNDS, 2021)
Procedure
The study was approved by the Nantes Health Ethics Group (22-03-910), and it was completed in accordance with Helsinki Declaration. Data from adults with NF1 were collected from neuropsychological assessments performed as part of their systematic follow-up between February 2020 and July 2021. Adult controls were recruited through advertisements and met at home. The assessment was spread over 2 half-days for all participants. Test administration and order were standardized according to the FSCB. Data on disease characteristics were collected from the expert dermatologist at the neurofibromatosis expert center, who were blinded to the results of the neuropsychological assessment. Regarding consent, patients were contacted after the follow-up assessment and informed of the study’s objectives. If they wished to participate, they were asked to return a signed consent form. An information letter was also sent to them, guaranteeing their anonymity. Adult controls provided their written informed consent before the tests began.
Statistical analysis
Statistical analyses were performed using SPSS software. As the normal distribution was not respected (Shapiro–Wilk test; all ps < .05), we carried out nonparametric tests. The sociodemographic comparability of the groups was verified using the Mann–Whitney test for age and education level, and the χ 2 test for gender. The means of the two groups for FSCB times and errors were compared using the Mann–Whitney test. Effect sizes were assessed using the pointbiserial correlation (r pb). An effect size below 0.1 was considered negligible, an r pb between 0.1 and 0.3 indicated a small effect size, an r pb between 0.3 and 0.5 indicated a medium effect size, and an r pb above 0.5 was interpreted as a large effect size. With regard to disease characteristics, means were compared between modes of inheritance using the Mann–Whitney test and between levels of severity and visibility using a Kruskal–Wallis test. The significance threshold was set at p = 0.05 for all analyses. A power analysis for each model evaluated was carried out using G*Power software in order to determine the minimum sample size required to accurately detect the true effect can be determined. For the main analyses, for the group comparisons performed using the Mann–Whitney test, considering an effect size of 0.8, a significance threshold of 0.05 and a desired power of 0.8, the minimum size of our samples had to be 19 participants each. For secondary analyses using the Kruskal–Wallis test, for severity (4 factors), the minimum overall sample size required was 24 patients. For visibility (3 factors), the minimum overall sample size required was 21 patients.
Results
Sample characteristics
The sociodemographic characteristics of our participants are set out in Table 2. Patients scored significantly lower than the control group on the Vocabulary subtest of the WAIS-IV (U = 302; p = 0.006), as well as on the self-report (U = 85; p = 0.016) and informant-report (U = 40; p < .001) scores of the RSAB. These metrics were controlled for in subsequent analysis. In terms of comorbidities, six of the 20 patients met the DSM-5 criteria for attention deficit hyperactivity disorder (ADHD; identified by semistructured interview and the administration of daily life questionnaires to the patient and a relative (RSAB); data not shown). None of the adults with NF1 had an autism spectrum disorder (ASD) comorbidity.
Table 2. Mean (standard deviation) sociodemographic and clinical characteristics of adults with NF1 and controls
Note. Significant results shown in bold. RSAB: Rating Scale of Attentional Behaviour; astandard scores.
Among the adults with NF1 (8 men, 12 women; mean age = 26.5 years, SD = 7.4 years), eight had a familial form, while 12 had a sporadic form. In terms of severity, three patients had a minimal form of the disease, seven a mild form, six a moderate form, and four a severe form, according to the Riccardi scale. For 10 patients, the disease was not very visible on the Ablon scale, for eight it was moderately visible, and for two it was very visible.
Social cognition
Table 3 provides the descriptive statistics (means and standard deviations) for the two groups on the social cognition tasks. Adults with NF1 performed significantly worse than controls on the Emotion Discrimination task (medium effect size) and had significantly slower response times for the Expressive Intensity Judgment and Emotion Discrimination tasks (medium effect sizes).
There were no significant differences on success scores for individual emotions. In terms of response times, adults with NF1 were significantly slower than controls for the emotions of happiness (U = 94.0; p = 0.004), fear (U = 113.0; p = 0.018), and surprise (U = 105.5; p = 0.018) in the Emotion Identification task (Figure 2). A nonsignificant trend was identified between the two groups for anger (U = 143, p = 0.12) and disgust (U = 135, p = 0.08). In the expressive intensity judgment task, NF1 adults were significantly slower than controls for the emotions of anger (U = 125.5; p = 0.045), happiness (U = 119.5; p = 0.03), and surprise (U = 89.0; p = 0.002). (Figure 3). It is important to note that the WAIS-IV processing speed index in the patient group was within the norm for the group as a whole (data not shown).
Figure 2. Mean response times per emotion in the French social cognition battery emotional identification task for adults with NF1 and controls. ** p < .01. * p < .05.
Figure 3. Mean response times per emotion in the french social cognition battery expressive intensity judgment task for adults with NF1 and controls. ** p < .01. * p < .05.
Analysis of the results for the social inference component of the FSCB revealed medium group effects for the first-order false belief success score of the ToM task, the seriousness score of the Moral and Conventional Judgments task (significantly higher for patient group on moral and conventional items), and the inappropriate score of the Social Situations task. No significant difference (variance equal to 0) was observed between the two groups on a physical inference item of the ToM task that was used to rule out a more general inference problem among participants.
Analyses of covariance were performed to ensure that there were no effects of language and attention on the statistical analyses comparing the two groups (no change in effects). The adults with NF1 with or without associated ADHD were comparable in sociodemographic terms, and no significant differences were observed between these two subgroups in the FSCB tests.
Influence of NF1 characteristics
Patients with different modes of inheritance, different levels of severity, and different levels of visibility were perfectly comparable on age, gender, and education level (all ps > .05), as well as on social cognition variables.
Table 3. Mean (standard deviation) raw scores of adults with NF1 and controls on french social cognition battery tests
Note. Significant results shown in bold. a variance equal to 0.
Discussion
The aim of the present study was to explore social cognition skills in adults with NF1 and the potential influence of disease characteristics on these skills.
With regard to emotion, we found significant differences between patients and controls on two of the three tasks, in terms of completion time. In addition, the success score of the Emotion Discrimination task was significantly lower in the NF1 group than in the control group. These results seem to validate the hypothesis of difficulties in emotional abilities in NF1, insofar as patients were seemingly able to identify emotional facial expressions, but required significantly more time to do so than controls. More detailed analysis of response times showed that they were longer for happiness, surprise, fear, and anger. These results are in line with those of previous studies conducted among adults with NF1, which identified impaired recognition of happiness, surprise, and anger (Pride et al., Reference Pride, Crawford, Payne and North2013, Reference Pride, Korgaonkar, Barton, Payne, Vucic and North2014). The greater differences for these emotions may be explained in part by the proximity of their facial expressions. This is particularly the case for the emotions of fear, happiness, and surprise, for which identification confusions have been described in several studies and/or by the authors who constructed the FSCB (Du & Martinez, Reference Du and Martinez2011; Ehrlé et al., Reference Ehrlé, Henry, Pesa and Bakchine2011; Jack et al., Reference Jack, Blais, Scheepers, Schyns and Caldara2009; Zhao et al., Reference Zhao, Yan, Chen, Zuo and Fu2013). Thus, the expressive proximity of these emotions made it harder for patients to tell them apart, explaining their significantly longer identification times for two of three emotional tasks compared with controls, as well as their significantly lower scores on the Emotion Discrimination task. However, it is necessary to qualify this explanatory hypothesis insofar as the control subjects also appeared to be slower on the same emotions than the NF1 patients. This could suggest that the items were of unequal difficulty depending on the emotions. Nevertheless, these results are consistent with previous findings indicating that certain emotions are processed more slowly than others, including in general population (Kirouac & Doré, Reference Kirouac and Doré1983; Kosonogov & Titova, Reference Kosonogov and Titova2019; Palermo & Coltheart, Reference Palermo and Coltheart2004). Moreover, there was a nonsignificant trend for the emotions of anger and disgust. This indicates that the observed differences may become more pronounced with larger samples of participants, and that the difficulties would then affect all the basic emotions. Consequently, these findings may indicate that the cognitive burden associated with identifying emotional facial expressions is greater in patients, resulting in a tendency to compensate for difficulties by prolonging completion times. Adults with NF1 tend to prioritize the quality of their responses over response time.
Regarding ToM, patients demonstrated a reduces performance on the attribution of first-order false beliefs, but did not differ from controls on either second-order false beliefs or missteps. These observations were unexpected, in the sense that the understanding of second-order false beliefs is predicated on success in identifying first-order false beliefs conditions (Duval et al., Reference Duval, Piolino, Bejanin, Laisney, Eustache and Desgranges2011). We can assume that a methodological bias influenced group comparisons. Indeed, although previous papers have addressed the issue of clinical sensitivity (Ehrlé et al., Reference Ehrlé, Henry, Pesa and Bakchine2011, Reference Ehrlé, Cavalier, Toulon, Delaby, Menu and Bakchine2015, Reference Ehrlé, Hody, Lecrique, Gury and Bakchine2020; Henry et al., Reference Henry, Tourbah, Chaunu, Rumbach, Montreuil and Bakchine2011, Reference Henry, Bakchine, Maarouf, Chaunu, Rumbach, Magnin, Tourbah and Montreuil2015, Reference Henry, Tourbah, Chaunu, Bakchine and Montreuil2017), this study represents a preliminary investigation. Further, more precise analyses of the psychometric validity of the tool should be conducted in order to confirm the strength of our measures. Furthermore, difficulties in first-order attributions, without any disruption of second-order attributions, could be explained by a double attribution error in the second condition, whereby participants did produce some correct answers, possibly fortuitously. Another hypothesis concerns the allocation of resources to the task. We can postulate that difficulties identified in the simple condition but not in the more complex one point to a failure to adjust resources in the simpler condition. Accordingly, the weaknesses identified in the social cognition tasks may ultimately have stemmed from executive problems, as patients with the neuropsychological phenotype of NF1 are now known to have a major executive impairment. These arguments suggest that ToM is relatively preserved in adults with NF1, thus invalidating our hypothesis. Nevertheless, these conclusions diverge from the results of a previous study that demonstrated partial ToM difficulties in adults with NF1 (Pride et al., Reference Pride, Korgaonkar, Barton, Payne, Vucic and North2014). These discrepancies may be explained by the fact that Pride et al. (Reference Pride, Korgaonkar, Barton, Payne, Vucic and North2014) relied on the interpretation of sarcasm to understand ToM, whereas we focused on the understanding of false beliefs and missteps. Another explanatory clue lies in the fact that the SI-M task used in Pride et al. (Reference Pride, Korgaonkar, Barton, Payne, Vucic and North2014)’s study also seems to involve the pragmatics of language. Although there is a consensus in the literature that pragmatics and ToM conceptually overlap, some authors urge caution, suggesting that pragmatic tasks cannot be regarded as a direct measures of ToM, in the sense that pragmatics and ToM are two distinct processes that cannot be considered subcomponents of each other (Bosco et al., Reference Bosco, Tirassa and Gabbatore2018).
Our results did not support the hypothesis that individuals with NF1 experience difficulties in moral reasoning. The patient and control groups did not differ on success on the FSCB Moral and Conventional Judgments task, with unethical and unconventional acts identified by both groups, and judged to be independent of authority. Nevertheless, patients rated the unethical and unconventional acts as more serious than controls did. Although this suggests that patients had no difficulty judging whether an act was moral or not, it nevertheless points to heightened sensitivity to morality in adults with NF1 compared with the general population. Moreover, this observation seems consistent with the particularly strong sense of justice/unfairness reported by patients at the clinic.
Alongside social conventions, other factors such as emotional reactions can also influence our judgment of the appropriateness or inappropriateness of a behavior. This is particularly the case in social situations where there are no moral issues at stake, but where certain characteristics may induce reactions of anger or irritation and thus be considered inappropriate (Blair & Cipolotti, Reference Blair and Cipolotti2000; Ehrlé et al., Reference Ehrlé, Henry, Pesa and Bakchine2011). In our study, patients had significantly more difficulty identifying inappropriate behaviors in social situations, and we observed weaker identification and recognition of anger, in line with previous research (Pride et al., Reference Pride, Crawford, Payne and North2013). More specifically, patients had no difficulty identifying normal situations, but failed to identify social violations, despite their thorough knowledge of social rules. We can therefore assume that their difficulties may have stemmed partly from weaker anger recognition that prevented them from correctly or fully identifying inappropriate behaviors in social situations.
We also investigated whether certain medical characteristics specific to NF1 influenced social cognition performances. With regard to mode of inheritance, none of the social cognition measures showed any effect of the familial or sporadic form of the disease, thus validating our hypothesis. These results corroborate those of previous studies conducted in a pediatric population, which found that quality of social life, social functioning, and ASD symptoms did not differ according to mode of inheritance (Garg et al., Reference Garg, Lehtonen, Huson, Emsley, Trump, Evans and Green2013; Klein-Tasman et al., Reference Klein-Tasman, Janke, Luo, Casnar, Hunter, Tonsgard, Trapane, van der Fluit and Kais2014; Plasschaert et al., Reference Plasschaert, Eylen, Descheemaeker, Noens, Legius and Steyaert2016). The same is true for severity levels, as no significant differences were found in socioemotional scores. These results are in line with the findings of a previous study that failed to find any impact of severity levels on social functioning in patients with NF1 (Chisholm et al., Reference Chisholm, Anderson, Pride, Malarbi, North and Payne2018), using only the Riccardi scale, without consideration of other neurological and neuropsychological impairments. However, for visibility, our hypothesis was refuted, as our analyses revealed no effect on patients’ overall social cognition scores. Albeit preliminary, these results suggest that increased disease severity does not negatively influence the social cognition abilities of patients with NF1. One of the factors contributing to these results is probably the early management of patients in childhood. For several years now in France, national healthcare recommendations have advocated early and regular neuropsychological follow-up of patients with NF1 (Centre de référence labellisé Neurofibromatoses, 2021). Furthermore, 35% of the patients included in the study had already undergone a neuropsychological assessment in our department, resulting in the formulation of treatment recommendations. This treatment, initiated in childhood (including before 2016 when the disorders identified required early treatment to limit their impact on daily life), could be considered a hypothesis to explain the blurring of differences between patients with varying levels of difficulty. However, this argument requires qualification, as it was not possible to ascertain the nature of the treatment actually received by each patient.
These results should be treated with caution, given that there were several methodological biases we were unable to control. First, the statistical power of the results was limited by the small size of our sample, although this appears reasonable given the rarity of the disease. More specifically, for the secondary analyses of our study (impact of disease characteristics), the patient sample seemed slightly insufficient for the power analyses performed. Furthermore, although all patients seen at the NF1 referral center were systematically referred to the learning disability center for a neuropsychological assessment, it is possible that some patients were not seen at the hospital for various reasons. This may have introduced a recruitment bias, limiting the generalizability of our results. Furthermore, the absence of ASD comorbidity and the very low proportion of patients with a highly visible form of NF1 reflect a probable recruitment bias in our clinical population. Moreover, although our study aimed to include adults aged between 18 and 59 years 11 months, in reality the age distribution and range of our final group deviated from this recruitment objective, as our patient sample was largely made up of young adults aged 18–43 years. The older third of the theoretical base sample was therefore not represented. This lack of representativeness prevents us from generalizing our results, especially given the heterogeneity of the patient profiles. Finally, it is evident that social cognition has not been comprehensively investigated, due to the complexity and the multiplicity of the processes it encompasses. Furthermore, the impact of NF1 patients’ social cognition disorders on their social functioning could not be investigated in this study, as social adaptation was not assessed. Consequently, further investigation into other concepts, such as pragmatic language, and their links with everyday social functioning, represents a promising avenue for future NF1 research.
It should be noted that the recruitment of participants was distributed over a period of 18 months, from February 2020 to July 2021. This raises the question of the impact of the COVID-19 pandemic on data collection. It appears that encounters with control subjects at home were relatively unaffected by the context. However, the number of cancelled appointments by patients in hospital was greater than before the COVID period, despite the introduction of measures designed to limit the risk of transmission in hospital (including compulsory wearing of a mask, the use of Plexiglas between the tester and the patient, and the requirement for a negative PCR test). This may have had an impact on the recruitment of patients for our study. Furthermore, the socioeconomic level of participants was determined solely on the basis of their level of education. This appears to be a relatively limited approach, particularly given that some of these variables may have changed following the pandemic, which could have a direct impact on patients’ ability to take part in the study. It would be beneficial to consider incorporating additional variables, such as family income into future studies to enhance the completeness of the data.
In conclusion, this exploratory study confirmed the difficulties of social cognition in adults with NF1. Results revealed emotional weaknesses identified in previous studies and, for the first time, difficulties in moral reasoning and social information processing. With regard to ToM, our results appeared to contradict those reported in a previous study, highlighting the possible impact of the methodology used to assess social cognition. No association was found between NF1 characteristics (mode of inheritance, severity, and visibility) and social cognition test results. While these preliminary results point to an additional possible explanation for the social difficulties encountered and reported by patients, future studies should include daily life questionnaires as a complementary measure, in order to identify the precise repercussions of social cognition impairments.
Funding statement
This article forms part of a doctoral dissertation. We warmly thank the Neurofibromatoses and Recklinghausen Association (OSE), as well as the University of Angers, which provided doctoral funding (A900201).
Competing interests
None.
