Volitional modification of brain activity in adolescents with Autism Spectrum Disorder: A Bayesian analysis of Slow Cortical Potential neurofeedback

Abstract

Autism spectrum disorder is (ASD) characterized by a persisting triad of impairments of social interaction, language as well as inflexible, stereotyped and ritualistic behaviors. Increasingly, scientific evidence suggests a neurobiological basis of these emotional, social and cognitive deficits in individuals with ASD. The aim of this randomized controlled brain self-regulation intervention study was to investigate whether the core symptomatology of ASD could be reduced via an electroencephalography (EEG) based brain self-regulation training of Slow Cortical Potentials (SCP). 41 male adolescents with ASD were recruited and allocated to a) an experimental group undergoing 24 sessions of EEG-based brain training (n₁ = 21), or to b) an active control group undergoing conventional treatment (n₂ = 20), that is, clinical counseling during a 3-months intervention period. We employed real-time neurofeedback training recorded from a fronto-central electrode intended to enable participants to volitionally regulate their brain activity. Core autistic symptomatology was measured at six time points during the intervention and analyzed with Bayesian multilevel approach to characterize changes in core symptomatology. Additional Bayesian models were formulated to describe the neural dynamics of the training process as indexed by SCP (time-domain) and power density (PSD, frequency-domain) measures. The analysis revealed a substantial improvement in the core symptomatology of ASD in the experimental group (reduction of 21.38 points on the Social Responsiveness Scale, SD = 5.29), which was slightly superior to that observed in the control group (evidence Ratio = 5.79). Changes in SCP manifested themselves as different trajectories depending on the different feedback conditions and tasks. Further, the model of PSD revealed a continuous decrease in delta power, parallel to an increase in alpha power. Most notably, a non-linear (quadratic) model turned out to be better at predicting the data than a linear model across all analyses. Taken together, our analyses suggest that behavioral and neural processes of change related to neurofeedback training are complex and non-linear. Moreover, they have implications for the design of future trials and training protocols.

Keywords: Adolescents; Autism Spectrum Disorder; Bayesian multilevel model; EEG Neurofeedback; Slow Cortical Potential training; Volitional brain activity modification.

Fig. 1

Study overview. Screening (including a diagnosis of ASD, IQ assessment and inclusion/exclusion criteria check), interventions in experimental and control condition, as well as pre-, post- and intervention-accompanying measures. In addition, on an exploratory basis, we assessed general mood, levels of motivation, concentration, fun, goal attainment, arousal and well-being (before every SCP training) as well as possible influences regarding treatment-related trainer and participant variables (4 times in the control- and 6 times in the experimental group via FERT (Klimesch et al., 2007); for details see SI (D), (E)). ADI-R: Diagnostic Interview for Autism-Revised ADOS-2: Diagnostic Observation Schedule for Autistic Disorders 2 FERT: Fragebogen zur Erfassung relevanter Therapiebedingungen (questionnaire for the assessment of therapy conditions) HAWIK IV: Hamburg-Wechsler-Intelligence Test for children and youth.

Fig. 2

The SCP neurofeedback setting. At the beginning of each trial, a triangle was displayed, specifying the polarity of the requested SCP shift of the upcoming regulation trial: a triangle pointing upwards required negative SCP shifts (increase of cortical activation), while a triangle downwards indicated required positive SCP shifts (inhibition of cortical activation). After baseline recording, the current SCP activity was displayed as an object (e.g., a fish or a moon) at the participants’ screen in real time and moved accordingly to the participants’ brain activity upwards (indicating an increase in cortical activation) or downwards (indicating a decrease in cortical activation). The participants should learn how to volitionally move the object up or down by controlling their SCP in the required polarity. All successful changes (i.e., declinations from baseline in the required polarity with duration of 2 consecutive sec in the last 4sec of each trial) were rewarded with the symbol of a sun after each trial and motivational feedback from trainers.

Fig. 3

Social Responsiveness Scale results from the quadratic model. The first row depicts the model’s mean predictions for each subscale along with 95% credible intervals (shaded regions). Solid black lines depict the empirical means on each occasion. The second row depicts the posteriors of the mean improvements in each group defined as the predicted mean score at $t_1$ minus the predicted mean score at $t_6$ of a given subscale. NFB = Experimental Neurofeedback Group.

Fig. 4

SCP neurofeedback results from the quadratic model. The first row depicts the quadratic model’s mean predictions vs. empirical means across all sessions for each Feedback condition × Task combination. The shaded regions indicate 95% posterior credible intervals. The marked dashed lines indicate the empirical means in each session, condition and task. The second row depicts the posteriors of differentiation (negativity – positivity) at the beginning of training, (D($t_1$), light green) and at the end of training (D(t₂₄), dark green). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 5

Task-related Power Spectral Density predictions from the quadratic models. The first row depicts the quadratic model’s mean predictions of mean PSD for the delta frequency band, the second row for the theta frequency band and the third row for the alpha frequency band. Predicted and empirical means are depicted across all sessions for each Feedback condition × Task combination. The shaded regions indicate 95% posterior credibility intervals. The marked dashed lines indicate the empirical means in each session, condition and task.