Differences in Neural Activation Associated with Visuomotor Sequence Learning in Autism

Background: Impairments in motor function are a consistent finding in children with autism.  It has been proposed that autism may be associated with abnormalities in neural processes underlying procedural learning leading to impaired development of motor, as well as social, skills.   The Serial Reaction Time Task (SRTT) is a commonly used method used to examine procedural learning, specifically visual-motor sequence learning.  Behavioral studies using SRTT in autism have produced mixed results.  It remains unclear whether neural processes underlying SRTT learning in autism are distinct from those of typically developing (TD) controls. 

Objectives: Examine for autism-associated differences in neural activation during visuo-motor sequence learning, as assessed using the SRTT.

Methods:  Two runs of fMRI while doing SRTT were recorded on 32 children, 10 children with autism (N males, mean ± SD Age) (8M, 10.7±1.2 years), and 22 TD controls (18M, 10.5±1.5 years).  Each run consisted of 7 blocks for a total of 168 trials; blocks 2-4 used 3 implicit 8-trial repeated sequences; blocks 1 and 5 were random, having 24 trials each.  There were 6 rest periods, 2 of 5 seconds at the beginning and end of the run, and 3 of 25 seconds between blocks.  Learning was assessed using repeated measures ANOVA (RM-ANOVA) to examine for a reaction time and accuracy differences across blocks. FMRI analyses were conducted using SPM5 with event related analyses modeling sequenced, random, and rest blocks.  Effect of sequence learning was examined using a contrast of activation of sequenced vs. random blocks.

Results: Behavioral analyses revealed that across groups there was a significant effect of sequenced blocks on percent correct responses (F=3.0, P=0.02) and mean reaction time (F=5.8, P< 0.001). There was no effect of diagnosis on reaction time or accuracy and no block x diagnosis effects.  FMRI analyses revealed a significant effect of diagnosis for the contrast of sequenced vs. random blocks, with children with autism showing more activation in the left medial temporal lobe (BA 21) compared with TD controls (cluster threshold corrected P = 0.004).

Conclusions:  Reaction time and accuracy relationships do not seem to be effected by diagnosis, though block effects persist.   Patterns of differences of sequence vs. random blocks revealing greater activation in the left medial temporal lobe suggests  children with autism may excessively rely on explicit/declarative mechanisms during visual motor sequence learning.