Working Memory Load-Dependent Activation Differences Between ASD and TD Children: An fNIRS Study

Background:  Autism Spectrum Disorder (ASD) is characterized by difficulties with social interaction, communication, and restricted and repetitive interests – however the disorder also manifests as disturbances in executive function. ASD subjects have been shown to have atypical brain activation on executive function tasks even where there are no differences in task performance. These studies have typically used fMRI, which is of limited use in children due to its sensitivity to head motion. In this study, we use fNIRS (functional Near-Infrared Spectroscopy), a method robust to head motion, to measure activation differences between ASD and typically-developing (TD) children during a working memory task. In addition, we examine whether brain activation correlates with executive function in daily-life using the Behavior Rating Inventory of Executive Function (BRIEF).

Objectives:  Determine whether activation in response to increasing working memory load differs between ASD and TD children, and whether activation differences correlate with real-world measures of executive function in daily life.

Methods:  Forty-one age- and IQ-matched subjects (19 ASD, 22 TD) of ages 7-15 and IQ >80 were given a 7-minute n-back task with loads of 0-, 1-, and 2-back. Each load condition contained 3 blocks of 9 consonant letter stimuli, with block sequence pseudorandomized. Optical data were recorded on a two-wavelength (690 and 830 nm) continuous-wave CW6 imaging system (TechEn, Inc., Milford, Massachusetts). The 60 optical channels covered the anterior and lateral frontal, temporal, parietal, and occipital cortices. Preprocessing: Raw signals were corrected for motion by spline interpolation, band-pass filtered, and converted to changes in deoxygenated hemoglobin concentration. Activation: Subject reaction times for correct target trials were z-scored and an average was taken for each condition. These values were then used to create a load-dependent regressor for each subject. A GLM was performed at each channel using NIRS-SPM and the t-statistic for load-dependent activation was computed, regressing out the covariates age, gender, and IQ. Group difference was statistically evaluated by 2-sample t-test. P-values were corrected using the false discovery rate (FDR) procedure. The channel with peak group difference in activation was then correlated with the Behavioral Regulation (BRI) and Metacognition (MI) indices from the BRIEF.

Results:  Group x load ANOVAs on task accuracy and reaction time both revealed a main effect of load (F_(2,74) = 9.144 , p < .0005; F_(2,74) = 7.616 , p < .001) but no significant main effect of group or group x load interaction. ASD subjects exhibited greater load-dependent activation than TD in left lateral prefrontal cortex (t₍₃₉₎ = 3.893 , p < .05; Figure 1). Activation in this region correlated with the BRI in ASD subjects (r₍₁₅₎ = .542 , p < .05) but not controls.

Conclusions:  ASD subjects showed greater load-dependent activation than TD subjects in left lateral PFC while completing the task with equivalent performance. Activation in this area correlated with daily-life executive functioning for ASD subjects. These data suggest that executive functions require greater effort from ASD subjects than their TD counterparts and these differences may extend to everyday tasks and behaviors.