David Crewther

　　澳大利亚 Swin burne 科技大学

　　Purpose:Autism is a developmental disorder characterized by impaired social interaction and communication，and by restricted and repetitive behaviours.Many individuals in normal life exhibit certain characteristics of Autism Spectrum Disorder.We were interested whether differences in visual perception and early visual processing(grouping，motion)shown by autistic children(Dakin&Frith，2005)would be demonstrated by adults of normal intelligence with high versus low AQ(autism quotient scale;Baron-Cohen，Wheelwright，Skinner，Martin，&Clubley，2001)，and whether such differences would be reflected in the electrophysiology of the magnocellular and parvocellular visual pathways。

　　Method:129 participants mainly from a young adult university population completed an online questionnaire(50 items)of the AQ scale and 15 low AQ(<11)and 14 high AQ(>20)were selected for further testing.Participants identified(reaction time/accuracy)the global or local form of locally or globally salient(and congruent or incongruent)Navon figures.In addition，infinite and limited(100 ms)dot lifetime motion coherence thresholds were measured.First and second order kernels of the non-linear visual evoked potential(VEP;VERIS multifocal system)using achromatic stimuli of low(24%)and high(96%)contrast，yielding waveforms indicative of magno and parvocellular function(Klistorner，Crewther，&Crewther，1997)were recorded

　　Results:A mixed design ANOVA with AQ score(high versus low)，target identity(global versus local)，congruency(congruent versus incongruent)and saliency(globally salient versus locally salient)demonstrated a significant Score*Target Identity*Saliency interaction(F(1，27)=9.20，p=.005)on the Navon figures task.Non-linear VEP grand mean average waves for first order and for the first slice of the second order kernel(K2.1)showed significant departures，with the magnocellular generated K2.1 showing manifest delay at high contrast in high AQ participants.Discriminant analysis(quadratic)with 1 input from the Navon Figure local salient task and 6 inputs from the VEP kernel peaks and latencies from the first and second order kernel slices were sufficient to correctly classify all of the High and Low AQ individuals(-2 log likelihood=0.421;minimum probability of classification for any individual=0.85)

　　Conclusions:The consequences of a delayed magnocellular VEP observed at high contrast for the high AQ group are a possible negation of the magnocellular advantage(Laycock，Crewther，&Crewther，2007)seen in normals，preventing any assistance from retroinjection of dorsal stream grouping information to area V1.This may explain the difficulty in identification of global forms in locally salient Navon figures.The correct discrimination of AQ level on the basis of largely physiological data lends support to the idea of a fundamental difference in sensory neural processing as the basis for autistic behaviour.