No Autistic Advantage in Inspection Time When Groups Are Matched Using the Raven's Progressive Matrices

Background: The Inspection Time (IT) Task measures visual processing speed and has good face validity as a method to estimate neural information processing time. It is strongly correlated with the Wechsler IQ, particularly performance IQ, in typical populations. However, conflicting results have been reported concerning IT in autism when using standard IQ estimation procedures to match group properties (Scheuffgen et al., 2000; Wallace et al., 2009). An alternative matching procedure utilizes a test of matrix reasoning ability, the Raven’s Progressive Matrices (RPM), a test for estimating general intelligence that is expected to be less affected by differences in verbal abilities than the Wechsler IQ. Autistics are known to achieve higher RPM scores than their Wechsler IQ predicts.

Objectives: Investigate the relationships among IT, RPM scores and Wechsler IQ measures in autistic and typically developing individuals.

Methods: We measured IT, matrix reasoning, and Wechsler IQ in 21 autistic and 28 typically developing participants (Wechsler FSIQ range: 78-126; age range: 14-34).  In the inspection time task, two vertical lines of different length were presented on a monitor for durations of 10-200 ms and then immediately masked by two irregular lines. Participants indicated the longer of the two lines by pressing one of two keys on a response box and no instructions as to speed of response were given. The stimulus duration was adaptively varied in a staircase psychophysical procedure. Between-group IT comparisons were made in subsamples matched for either FSIQ or RPM percentile. We used a mixed effects model with group as a between-subjects factor and subject as a random factor.

Results: Using the FSIQ as a group matching variable, autistics had shorter inspection times (autistic mean IT = 88.4 ms and typical mean IT = 113.4 ms; p = .042).  In contrast, using RPM percentile as the group matching variable yielded different results, with autistic and typical samples showing no IT differences. Examination of the correlations among the measures provided an explanation for the different results obtained with the different group matching strategies. For autistics, there was a strong correlation between RPM and IT (r=-.79, p<.001) but not between FSIQ and IT (r=-.36, p=.114). In the typical sample RPM percentile and IT (r=-.57, p=.002) were correlated more strongly than FSIQ and IT (r=-.21, p=.281).

Conclusions: Because of the observed shared variance between IT and matrix reasoning, it is clear that some care should be taken in selecting matching variables for studies of cognitive abilities in autism. The apparent advantage observed in autistics for IT task may actually result from using matching variables that share processing mechanisms with target task.