Functional Neuroanatomical Changes Induced by Mu-Based Neurofeedback Training in Children on the Autism Spectrum

Background:  Autism Spectrum Disorders (ASD) may arise from atypical anatomical and functional connections and therefore have been characterized as a ‘disconnection syndrome’. Impaired connectivity may lead to desynchronization and ineffective intra- and interhemispheric communication in neural circuits affecting higher order cognitive processes.  While no single explanation can account for the ASD profile, converging evidence implicates the human mirror neuron system (MNS).  Studies from our laboratory have shown that ASD individuals exhibit normal EEG mu rhythm suppression for self-generated movement but fail to suppress during observation of movement compared to typically developing (TD) controls. On the other hand, suppression is normal if the actors being observed are familiar, suggesting that the MNS is not entirely broken. We have shown that significant improvement occurs in social engagement and related behaviors, as well as in the electrophysiology of ASD children following neurofeedback training focused on the mu-rhythm.

Objectives: The present study tested whether functional and structural neuroanatomical changes occur after 20 weeks of mu-based neurofeedback training.  

Methods:  Neurofeedback training is an operant conditioning task in which trainees learn to control mu rhythm (8-13 Hz) power at electrode site C4, over the sensorimotor cortex in the right hemisphere.  Games and movies on a computer reward increased mu-power and decreased muscle activity.  All participants complete 30 hours of this training (45 min/session x 2 sessions/week x 20 weeks).  Prior to and again immediately following training, participants (7 ASD and 8 TD, ages 8-17) underwent fMRI scans that included the following protocols: resting state fMRI (6 min), 3 fMRI runs of a task that involved imitation and observation of object-oriented finger movements (total of 15 min), anatomical (5 min), and diffusion tensor imaging (10 min).  Contrary to imitation tasks previously used by Iacoboni (1999) and Williams (2006), the imitation task in our study was object-oriented (pressing buttons on a button-box). Diffusion tensor imaging data were collected to assess white matter changes associated with neurofeedback training in pathways connecting areas of the MNS. 

Results:  Before training, greater activation occurred in regions of interest related to MNS in TD compared to ASD during object-oriented imitation and observation. These areas of differential activation included left inferior frontal gyrus (IFG) and bilateral inferior parietal lobules.  Abnormal resting state functional connectivity (both under- and over-connectivity) between MNS regions of interest was also seen in ASD compared to TD groups.  Following mu neurofeedback training, ASD children showed increased activations in IFG and other relevant MNS areas, as well as normalization of functional connectivity in MNS circuits.  

Conclusions:  These preliminary data indicate plasticity within the mirror neuron system occurs in response to mu-based neurofeedback training in ASD. Both activation and connectivity measures were found to normalize with training.