Autism spectrum disorders (ASD) and psychotic disorders, such as schizophrenia, both represent severely disabling neurodevelopmental disorders (Goldstein et al. 2002
), with marked impairments in cognition and social functioning (Couture et al. 2010
; Sasson et al. 2011
). For ASD, the first behavioral problems typically occur during early childhood, while psychotic disorders are primarily diagnosed in late adolescence and young adulthood. Despite some symptomatic overlap, both psychiatric classifications are largely characterized by differential behavioral phenotypes and appear mutually exclusive in diagnostic manuals such as the Diagnostic Statistical Manual of Mental Disorders
(5th edition; DSM5; American Psychiatric Association 2013
) and the International Classification of Diseases
(10th edition; ICD-10; World Health Organization 1993
). Interestingly though, there is substantial evidence indicating that children diagnosed with ASD have an increased prevalence of psychotic disorders later in life and that psychotic disorders are associated with increased rates of ASD (Chisholm et al. 2015
; Selten et al. 2015
In recent years there has been a renewed interest in specific overlapping autistic and psychotic symptoms, in clinical and non-clinical samples, partially guiding the ongoing quest for differential markers and potential risk factors for psychosis in patients with ASD (Abu-Akel et al. 2016
; Barneveld et al. 2011
; Brosnan et al. 2014
; Chung et al. 2014
; Crespi et al. 2016
; Eack et al. 2013
; Sasson et al. 2016
). One overlapping symptom that has received relatively little attention is formal thought disorder (FTD). FTD refers to a disruption in the flow of thought, as observed by disorganized speech. Although speech is an indirect measure of thought, “language serves essentially for the expression of thought
”, as phrased by the famous linguist Noam Chomsky (Chomsky et al. 1979
), whose extensive work in this field implies that an understanding of the rules of a language throws light on the principles that regulate human thought.
FTD represents a hallmark feature of schizophrenia and is also characteristic of childhood onset schizophrenia, schizotypal personality disorder (Caplan 1994a
), and predictive of psychosis in adolescents at clinical high-risk (Bearden et al. 2011
). Likewise, speech disturbances, such as pragmatic language impairments, are a common feature in ASD. Obviously, it requires a minimal amount of speech in order to tap into the organization of thought in terms of logic and coherence. Indeed, when assessing FTD in verbal, high-functioning individuals with ASD there is some evidence of increased levels of FTD. Earlier studies in small adult samples of high-functioning individuals reported that, compared to adults with schizophrenia, those with ASD demonstrated more ‘negative’ thought disorder, i.e. poverty of (content of) speech, but not so much ‘positive’ features, i.e. illogicality and derailment (Dykens et al. 1991
; Rumsey et al. 1986
). More recently, elevated rates of FTD have also been reported in children with high-functioning ASD (Solomon et al. 2008
; van der Gaag et al. 2005
). These studies did report elevated rates of ‘positive’ FTDs (i.e. ‘illogical thinking’ and ‘loose associations’) in ASD compared to typically developing children. This is relevant, because this may point to disorganization of thought in a way that may predispose to symptoms that are typically observed in individuals with psychotic disorders.
There currently is no evidence suggesting that FTD is indicative or predictive of a subsequent psychotic episode in children with high-functioning ASD (Eussen et al. 2015
; van der Gaag et al. 2005
), although this has never been investigated in a direct manner. However, one long-term clinical follow-up study of a group of 55 ASD children meeting criteria for multiple complex developmental disorder (MCDD), a descriptive ASD subtype marked specifically by deregulation of thought and emotions (Buitelaar and van der Gaag 1998
; Sprong et al. 2008
; van der Gaag et al. 1995
), reported that approximately 70% of participants met criteria for schizophrenia spectrum disorder in adolescence or adulthood (van Engeland and van der Gaag 1994
). As such, it remains uncertain to what extent early observation of FTD may contribute to the development of psychotic disorders in later life.
In addition to a clear link with impaired semantic processing skills, there is a seemingly strong association between impaired executive functioning and FTD in patients with schizophrenia (Docherty 2012
; Kerns and Berenbaum 2002
). The term ‘executive functions’ refers to a set of cognitive processes associated with the control of thoughts and actions (Bunge and Souza 2009
). Executive functions include, but are not limited to, cognitive abilities such as response inhibition, working memory/updating, and set shifting (Friedman and Miyake 2016
; Miyake et al. 2000
). It is well established that executive functions are impaired along the full width of the psychosis spectrum (e.g. Bora and Murray 2014
; Giakoumaki 2012
; Ziermans 2013
), although results have been mixed concerning their added use for predicting psychotic onset on an individual level (Fusar-Poli et al. 2012
; Lin et al. 2013
; Metzler et al. 2016
; Ziermans et al. 2014
). Executive dysfunction has also been historically linked to ASD (Pennington and Ozonoff 1996
) and impairments have been widely reported (for a review see Russo et al. 2007
), albeit within the context of large individual and age-dependent differences (Pellicano 2010
; van den Bergh et al. 2014
). Such individual differences in ASD may also partially account for the observed symptomatic overlap with psychotic disorders, but only a limited amount of studies have addressed this issue.
Only one pilot-study has previously investigated FTD in relation to executive functioning in ASD (Solomon et al. 2008
). Solomon and colleagues compared 17 adolescents with high-functioning ASD to 21 matched controls on objective rating scales for FTD and examined correlations with one executive control task (Preparing to Overcome Prepotency; POP-task; Barber and Carter 2005
), which measures inhibition of a prepotent response. They found that only one type of FTD (illogical thinking) was specifically correlated to response inhibition with borderline significance. However, sample sizes were small and this study focused specifically on inhibition, which limits the ability to firmly establish which aspect of EF is most relevant for understanding FTD.
The first goal of the current study was to compare the prevalence of FTD, as measured by both objective ratings and subjective self-reports, between a substantial group of high-functioning children and adolescents with ASD and their typically developing controls (TDC), matched for age, gender and (verbal) IQ. Based on two previous reports in smaller samples, we expected that children and adolescents with ASD would show higher levels of FTD, in particular for ratings of illogical thinking and loose associations (Solomon et al. 2008
; van der Gaag et al. 2005
). Second, we aimed to determine whether level of FTD was impacted by cognitive performance on three core executive functions (response inhibition, working memory and cognitive flexibility) in high-functioning ASD. Establishing the relative contribution of executive functions to FTD, in the context of relatively preserved verbal skills, can potentially lead to improved early identification of cognitive risk factors for the development of psychotic symptoms in ASD, as well as provide new incentive for fine-tuning research into the underlying neurodevelopment of both types of disorders. We expected to find multiple associations between executive functions and FTD parameters and, more specifically, that response inhibition measures would explain a significant amount of variance in FTD (Barneveld et al. 2013
; Solomon et al. 2008
The findings in the current study suggest that children and adolescents with high-functioning ASD experience elevated levels of FTD, both objectively and subjectively, even in the context of intact (verbal) intellectual functioning. This corroborates previous accounts of increased rates of FTD in children and adolescents with high-functioning ASD (Solomon et al. 2008
; van der Gaag et al. 2005
). Few studies have investigated putative cognitive mechanisms driving these observed difficulties in the organization of thought and speech in ASD. Executive functions represent a set of interrelated cognitive skills that allow people to exercise a certain amount of control over their thoughts. These cognitive functions are often impaired in ASD and sometimes linked to FTD in schizophrenia spectrum disorders (Barrera et al. 2005
; Docherty 2012
). As such, we hypothesized that impaired executive functioning, in particular inhibitory control, would predict levels of FTD in ASD as well. Our study showed that, when evaluating multiple executive functions, verbal working memory was the single one associated with FTD.
Increased presence of FTD may or may not predispose adolescents for psychotic disorders. Evidence from a clinical high-risk population (CHR), presumably without ASD diagnosis, suggests that higher ratings for illogical thinking and poverty of content can help predict the onset of psychosis and level of social functioning approximately one year after intake (Bearden et al. 2011
). The only known follow-up study that investigated predictive validity of FTD in high-functioning ASD reported that illogical thinking in young children did not predict prodromal symptoms in adolescence 7 years later, and instead was more indicative of ASD symptom severity (Eussen et al. 2015
). However, the outcome measure consisted of a screening questionnaire for prodromal symptoms, assessed in children aged 12–20 years, so before the peak age of a first psychotic episode, and only 2 of 32 participants scoring above the screening threshold met formal criteria for CHR. Despite a few studies reporting on elevated levels of CHR symptoms in ASD and vice versa (Eussen et al. 2015
; Solomon et al. 2011
; Sprong et al. 2008
), study samples are typically not screened concurrently for both conditions, and therefore very little is known about the predictive validity of prodromal symptoms in ASD. One 6-year follow-up study (de Wit et al. 2014
) showed that out of 17 adolescents with ASD diagnosis and clinical high-risk, one individual (6%) had become psychotic, 5 (29%) were still considered at-risk, and 11 (65%) had remitted. Clearly additional follow-up studies in larger ASD/high-risk cohorts are required to address whether FTD constitutes a true risk factor for psychotic disorders in ASD.
Despite a global increase in overall FTD in our study, specific K-FTDS subtypes such as ‘Loose Associations’ and ‘Illogical Thinking’ did not differentiate between individuals with ASD and controls, which is inconsistent with the large effect sizes reported for these measures in previous studies (Solomon et al. 2008
; van der Gaag et al. 2005
). The only subtype that differed between groups was ‘Incoherence’, which denotes utterances that are difficult to understand due to insufficient organization. However, all effect sizes were small. A closer look at K-FTDS ratings across studies suggests that the average ratings in our study were quite low compared to the study by van der Gaag et al., possibly due to inclusion of a more mildly affected, slightly older ASD group and a lower utterance count. In terms of age, our study sample was similar to the Solomon et al. sample and FTD ratings for ASD were also highly comparable. However, FTD ratings for the control group in the Solomon study were virtually absent, whereas our control group did show some variation across FTD subtypes, and may therefore have been representative of a broader population. In addition, our sample sizes for both groups were roughly two-to-three times larger than in the two previously conducted studies, although the van der Gaag study also included an additional clinical comparison group diagnosed with MCDD. Interestingly, the MCDD group did not differ from the other ASD group on K-FTDS ratings.
To complement the objective measurement of FTDs, a subjective measure of FTD was also included in this study: the Odd Speech subscale of the Schizotypal Personality Questionnaire. For this measure we did find a significant group difference in the expected direction. Although the use of self-report questionnaires in individuals with ASD is sometimes scrutinized, the Odd Speech items are rather concrete and ask for both personal and observer qualifications of the subject’s speech difficulties. The fact that they are worded so that subjects also report on external corroboration of these symptoms probably indicates why the subscale can be measured quite reliably (Raine 1991
), and why it has previously been used as a measure of FTD in patients with schizophrenia (Badcock et al. 2011
). Training and rating procedures of K-FTDS are very time consuming and difficult to complete during clinical assessment (de Bruin et al. 2007
), and therefore it is worth investigating alternative FTD measures that require less time and effort and may be more suitable for application in clinical or research settings where limited resources are available. Even though continuous measures for Odd Speech and K-FTDS were not significantly correlated, the binary variables (based on clinical cut-offs) were (Odd Speech − FTD Total: ϕ
= 0.38, p
≡ .01). Although this suggests that both measures are only moderately tapping into the same construct, it is not uncommon for informant- and performance-based tasks measuring similar constructs to show limited correlations (Toplak et al. 2013
). Clearly, different types of rater-bias and test impurity can dampen the strength of these monotrait-heteromethod correlations, possibly because one or several mediating variables are unaccounted for. However, both methods may also provide complementary information on the same underlying latent construct. Therefore, our first recommendation is to further investigate the reliability and validity of the Odd Speech subscale as a proxy for FTD in a multitrait-multimethod matrix and second, to establish whether it can explain additional variance in prediction models of psychosis, for example in CHR samples.
Despite robust findings of executive dysfunction in ASD, the literature on this topic is vast and riddled with conflicting findings. For example, it has been claimed that response inhibition is the only core executive function not impaired in ASD (Russo et al. 2007
). However, a recent meta-analysis on prepotent response inhibition and interference control in ASD concluded that the evidence suggests otherwise (Geurts et al. 2014
). Substantial inconsistencies have also been highlighted for working memory and cognitive flexibility performance in ASD (de Vries and Geurts 2014
; de Vries et al. 2015
), although rigorous meta-analyses for these constructs are currently lacking in the literature. Additionally, executive dysfunctions tend to have limited discriminative value across psychiatric classifications. Enriching between-group analyses with within-group associations may therefore be better suited to help identify specific associations between cognitive markers and heterogeneous behavioral phenotypes.
By applying this strategy, we were able to detect an association between executive functioning and FTD within the ASD group. Lower verbal working memory performance significantly predicted higher levels of FTD in both linear and logistic models and for both subjective and objective FTD in the latter. This suggests that individuals with ASD who experience verbal working memory difficulties may be particularly vulnerable for developing subsequent FTD symptoms. The task used (CELF-NR) is a digit span task, which (along with other distractibility tasks) has previously been associated with FTD in children with ADHD, but not in children with schizophrenia (Caplan et al. 2001
). Given the high comorbidity of ADHD and ASD symptoms in general, it is possible that the association was mostly reflecting increased distractibility in some of our individuals with ASD. This would subsequently limit the capacity to encode and retrieve verbal information in/from working memory and thereby reduce reproduction of the presented stories in the story game, for example. However, the absence of similar associations for visuospatial working memory, as well as for the other non-verbal executive functioning parameters, and the relative preserved verbal skills which were not correlated to FTD in our ASD sample, all strengthen the notion that executive control over verbal processing is key to understanding FTD in ASD. Furthermore, Docherty (Docherty 2012
) also found that verbal working memory (digit span task) predicted FTD in adults with schizophrenia. Although this does not directly implicate an increased risk for schizophrenia in ASD individuals with FTD, it could entail that the combination of clinical levels of FTD and verbal working memory problems poses a potential risk factor for psychotic episodes in ASD.
Two known studies have also directly addressed relations between executive functioning and FTD in ASD. Both indicated that prepotent response inhibition in ASD was significantly associated with FTD, respectively for K-FTDS—Illogical thinking (Solomon et al. 2008
) and SPQ – Disorganization (= Odd Speech + Eccentric Behavior) (Barneveld et al. 2013
). However, these studies consisted of smaller samples and correlational analyses were conducted with similar but slightly different parameters for only one executive functioning measure. Given the hypothesized dependent nature of FTD, regression analyses can provide better clues as to the relative impact of executive functions on FTD.
Two main limitations of our study need to be highlighted. Although we were able to increase sample size and include additional cognitive parameters and analyses compared to previous FTD studies in ASD samples, our study was still too limited to include additional predictors from other relevant neurocognitive domains, such as attention and language, which we recommend for future studies. An additional limitation is the use of cross-sectional data, which prohibits any inferences about the causal nature of any associations under investigation.
To conclude, the current study aimed to investigate relations between executive functions and FTD in children and adolescents with ASD and matched typically developing controls. In sum, we found evidence for an increased prevalence of FTD and a significant negative association between verbal working memory skills and FTD, which may overrule the potential impact of response inhibition or cognitive flexibility on disorganized speech. We therefore suggest that poor verbal working memory skills may predispose some children and adolescents with autism to develop thought disorder and advise researchers in the field to shift their focus from solely on executive control to include the role of executive verbal processing skills in relation to FTD in this target population.