Recent developments in neuroimaging methods15
have demonstrated aberrant brain connectivity15,16,17,18,19,20
and/or aberrant lateralisation9,10,20,21
in individuals with ASD. However, our study is the first that focuses on the brain functional connectivity and its cognitive correlates in young children with ASD under conscious conditions.
Previous functional magnetic resonance imaging (fMRI) studies have reported that the neural substrates for visual reasoning exist in a widely distributed fronto-parietal network22,23,24,25,26
. Intriguingly, one recent study of gifted and control adolescents demonstrated that the superiority in visual reasoning ability was driven not by additional cortical activation but by increased activation in the posterior parietal cortex, including the superior parietal lobule and the right intraparietal sulcus24
. It can therefore be supposed that more efficient processing during visual reasoning tasks is associated with more localised (or independent) brain activity within posterior brain regions. Consistent with this assumption, we did not observed positive significant correlations between higher visual reasoning ability and anterior-posterior brain connectivity in either TD children or in children with ASD.
Using fMRI in healthy adult participants, a previous study focused separately on the neural correlates of visual analytic and figural reasoning tasks. This previous study demonstrated that analytic reasoning yielded (i) greater activation in the left than in the right hemisphere, and (ii) greater anterior than posterior activation. In contrast, figural reasoning yielded (i) greater activation in the right than in the left hemisphere and (ii) greater posterior than anterior activation25
. The right posterior brain region is therefore crucial for figural reasoning abilities rather than analytic reasoning abilities. We find that rightward lateralisation of brain connectivity in posterior brain regions contributes to visual pattern reasoning in young children with ASD but not in TD children. This result suggests that individuals with ASD, who exhibit well-documented preservation in visual reasoning1,2,3,4,26
, could use figure perceptual strengths during any visual reasoning task.
A recent fMRI study demonstrated the neural bases of the solving of visual reasoning task in adults and adolescents with ASD26
. These authors demonstrated that subjects with ASD displayed relatively increased task-related activity in the posterior region (e.g., the occipital cortex) and decreased activity in the anterior region (e.g., the prefrontal cortex) during a visual reasoning task. Interestingly, these authors observed a trend towards increased activity in the right inferior parietal cortex in individuals with ASD during visual reasoning tasks. It has also been reported that the cognitive strategies that are adopted by individuals with ASD are different from those that are used by TD subjects during other visual tasks. For example, subjects with ASD exhibit right lateralised brain activity during a visual search task27
, greater posterior brain activity during an embedded figures task28,29
and right lateralised brain connectivity during an n-back visual working memory task21
. In conjunction with our results, it can be postulated that a functional preference for the right-posterior brain may play a prominent role during various visual tasks in individuals with ASD.
Language and visual search tasks are generally thought to activate the left and the right hemisphere, respectively. In addition, recent studies have reported that the development of language5,30
and visual search abilities30
is accompanied by the leftward and rightward lateralisation of brain activities, respectively. With regard to visual letter perception, a recent study demonstrated that invariant representations of letter identities are generated in the visual word form area (VWFA), which can be reproducibly identified in the left occipito-temporal sulcus31
. The VWFA then projects to structures that are involved in phonological or lexico-semantic processes in the left hemispheric language area31,32,33
. In contrast, a controlled, longitudinal training study of young non-reading kindergarten children demonstrated the initiation of both the right and left occipito-temporal cortex sensitivity to printed letters34
. This result suggests that the right-hemispheric analogue of the VWF (R-VWFA) also contributes to the ability of young children to read letters. Intriguingly, in cases of lesion of the VWF in adults, the R-VWFA was reported to participate in residual reading abilities (e.g., letter-by-letter reading)31
. Letter-by-letter reading implies that the normal ability to identify letter strings in a quasi-parallel fashion is lost. Patients who read letter-by-letter can decode the letters but fail to recognise letter strings as whole words using lexico-semantic processes. Given that right posterior brain regions, including the R-VWFA, do not have a practical role in higher level language processes but play a certain role in letter decoding, the observed right-posteriorly weighted brain function in individuals with ASD may be unsuitable for practical language processes. Consistent with this assumption, the group of children with ASD with normal fluid intelligence in the present study tended to exhibit a preserved ability on simple reading/decoding tasks but a significantly lower ability in verbal reasoning tasks. Furthermore, language impairment is one of the key features of ASD, and the failure to develop sophisticated language is one of the earliest signs of ASD35,36
Theta-band oscillations have been believed to represent ongoing cognitive processes during memory tasks7
. In addition, long-distance connectivity in the right hemisphere via theta band oscillation is believed to play a significant role in ongoing cognitive processes, especially during visual memory37,38,39
. In the present study, TD children with right lateralised anterior-posterior long-distant connectivity via theta band oscillations may have had a certain degree of cognitive strength in figural memory. Therefore, such individuals may exhibit higher letter reading performances. In contrast, in children with ASD, the strength of the anterior-posterior connectivity via theta band oscillations was not correlated with letter-reading ability. This result was consistent with the aforementioned assumption that individuals with ASD tend to make use of right-posterior regions (i.e., more localized region) during various visual processes.
Gamma-frequency oscillations between neural networks are essential for cortical information processing7,40
, and oscillations in posterior regions are believed to represent ongoing cognitive processes during visual perception and attention41,42,43,44,45,46,47,48
. Accordingly, the topographical distribution of the gamma band network in posterior regions during the watching of a video programme may represent the individual features of the brain network that are involved in the visual processes of daily life. In the present study, direct comparison between the ASD and TD groups demonstrated that the brain connectivity via gamma oscillations is higher in children with ASD in posterior region (i.e., temporo-occipital connection in the right hemisphere) (). Furthermore, gamma band oscillation-induced right-posteriorly weighted brain connectivity was associated with visual reasoning () and letter-reading ability () in children with ASD but not in TD children. This result was consistent with the aforementioned assumption that individuals with ASD tend to make use of right-posterior regions during various visual processes.
Although MEG is not the only child-friendly imaging technology, it is beneficial in studies that investigate the laterality of cortical oscillations. This benefit lies in the fact that magnetic fields that are generated unilaterally tend to reflect to sensors in the ipsilateral hemisphere. Furthermore, our new device (a custom-made MEG system for children) simultaneously recorded the right and the left hemispheres in young children (). This result would have been difficult to achieve had a conventional, adult-sized MEG system () been used given the small head size of children relative to the sensor array. This study is the first to indicate that brain functional connectivity in the right hemisphere and/or rightward lateralisation in posterior brain regions contribute to natural reading and/or visual pattern reasoning abilities in young children with ASD. Our results suggest that the observed functional connectivity in the right hemisphere and the rightward lateralisation of intrahemisheric connectivity in posterior brain regions was associated with the preserved visual task performance in children with ASD. In addition, this contribution of rightward lateralisation to cognitive performance diminished with age, suggesting a delayed maturation of certain unspecified neural pathways in children with ASD. We emphasise that it is time for a new era of insight into the brains of younger children with ASD under conscious conditions.
Converging evidence suggests that the properties of gamma oscillations are altered in ASD during information processing49,50,51,52
. In children with ASD (aged 3 to 8 years), one previous study demonstrated that an excess of EEG gamma band oscillations during sustained visual attention is directly related to the degree of developmental delay49
. This altered gamma-band activity that was observed in young children with ASD may result from disturbances in GABAergic or glutamatergic mediator systems, which are critically important to generate this type of oscillation53
, and suggested that changes in the excitation/inhibition balance may be a pervasive feature of cortical networks in ASD from a young age54
. There are two possibilities to explain our results (an altered contribution of brain connectivity to cognitive performance in ASD). The first explanation is that we observed an ongoing aberrant excitation/inhibition balance, which brought on the diversified cognitive ability in young children with ASD. The second explanation is that an aberrant excitation/inhibition balance during the foetal period results in aberrant prenatal and perinatal development that leads to lasting alterations in neural networks.
The present study had several limitations. First, we could not evaluate the degree to which the subjects attended to the auditory or visual information in the video programmes. The children with ASD might have attended to the visual information rather than to the narrative sound information. Differences in these modality-dependent preferences may be reflected in the functional brain connectivity. Second, we eliminated any contaminated MEG data, such as when clear ocular movement occurred. However, differences in the fine ocular movements between the children with ASD and the TD children could have confounded the results when frequent saccades occurred, as may have happened during the watching of the video programme. Third, we recorded the head positions of the subjects using video monitors during the MEG recordings, and we eliminated any MEG data by visual inspection for coherence analysis if the head position of the subject had obviously moved from its starting position. Further study using a quantification algorithm for the head movement will provide more reliable evidence. Fourth, more than 40-sec periods were recommended when computing coherence55
; however, after eliminating the contaminated data in the present study, the recording period that was accepted for analysis tended to be short (a minimum of 35 sec). In addition, the reliability of coherence in the gamma band is reported to be lower than that in the alpha band55
, and these coherences could be significantly affected by the mental state of the subjects56
. Further studies that employ longer periods and attention-controlled conditions will provide more reliable evidence, although these conditions will be difficult to achieve in conscious young children. Fifth, given that young children were examined in the present study, we were unable to obtain brain structural information on which to superimpose the coordinate systems of the source-estimated MEG signals. This limitation was encountered because it is troublesome, especially when studying young children, to perform additional MR imaging. We therefore performed sensor-level analysis instead of voxel-based analysis. To deduce the anatomical locations of the signal sources and to draw definitive anatomical conclusions, it will be necessary to perform further studies that use source localisation methods in combination with individual brain structural information.