FACE Processing IN Children with AUTISM Through Dense-ARRAY EEG Recording

Background:

Face processing in autism has been widely investigated during the last decades, nonetheless many open questions still stay open. The special role of human face in human interaction since the very early epochs of life along with the easy to notice disturbance in ASD affected individuals in handling face to face contact and facial expressions are the main reasons of such a wide interest. Caveats in this kind of studies are mainly the usually limited cooperation by tested ASD subjects and the ongoing maturation of face processing specific ERP peaks which appears to keep going until late adolescence. 

Objectives:  

Our aim is to better characterize ERPs in response to stimuli consisting of emotional and neutral faces as well as matched non-face objects in a sample of children diagnosed with ASD who underwent an extensive neuropsychological and cognitive assessments.

Methods:

In order to overcome the cited common issues we employed an EGI dense-array EEG system (GES 300, 128 electrodes nets) and restricted the study to a relatively narrow age range. The hi-density approach was chosen mainly because it is a fast and comfortable way of mounting so many electrodes that provide two main benefits: higher topographical accuracy of ERP peaks and in particular a very reliable estimate of average used for re-referencing post processing. The experiment consisted of 4 blocks of 5 minutes each consisting  in random presentations of 4 kinds ot stimuli (neutral, sad and happy faces, trees) and a fifth kind of stimulus that was used only to collects answers by a button press to keep the attention focused on the experiment. For every class of stimuli, a minimum of 50 artifacts free repetitions (80 when full run)  was obtained for every subject. Every epoch consisted of a 500-1500 msecs displaing a fixation cross, followed by a stimulus for 850 msecs. Stimuli were converted to black and white pictures and standardized for luminance and contrast. We used a widely used standarzide database of facial expressions (ref NimStim). Raw data was  1-30 Hz filtered, segmented, re-referenced to average and averaged across categories for each subject. Artifacts detection and bad channels replacement where  applied with both automatic and manual procedures.

Results:  

We measured latency and amplitude of the occipito-temporal N170 and of the preceding and following peaks. In both groups we found a differential activation in response to faces vs trees stimuli: in the control group we found a big difference in amplitude between the two classes of stimuli, difference that was present but however smaller for the ASD group. In other words N170 peaks were significantly more negative for faces and more positive for trees in the typical developing group. Regarding latency, while the control group exhibited similar latency for both stimulus categories, the faces stimuli in the ASD group evoked a later latency peak. Moreover in several ASD subjects the N170 peak appeared bifid, as previously described. 

Conclusions:  

This work adds evidence to the growing data on a possible involvement of early face processing steps in the development of ASD.