Objectives: We aimed to replicate and extend the Ozonoff study by comparing person-administered (Person Only, PO), and computer-administered (Computer Only, CO) versions of the WCST and by adding a third condition: we videotaped a person administering the task and showed the administration on computer (Person-Computer, PC). In accordance with Ozonoff’s hypothesis, we expected that performance on the PC version would fall between the CO condition (best) and the PO condition (worst) for autism relative to controls.
Methods: We tested 45 children and adolescents diagnosed with autism spectrum disorders (ASD), 15 in each test condition. ASD was diagnosed by an expert rater based on information from the standardized ADOS-G clinician observation and SCQ parent checklist; Verbal and Full Scale IQ were above 80 for all participants. An age-, sex-, and IQ-matched comparison group (n = 45) also completed the task. The PO version was administered using standard instructions (Heaton, 1993). The CO and PC conditions were presented using our own E-Prime-based software program. The card stimuli on the CO version were identical to those used in the actual card set, and the young adult male shown in the PC version used the actual cards. We made one modification to the CO task compared to standard commercially-available versions: in order to simulate the lag time that accompanies the human administrator reaching for the card deck and raising it to show the participant, we added a short (2 second) video of a point-light display of a rotating circle or square in-between each trial. Total time from the end of one trial to the possibility of making the next choice was about 4 seconds in all three conditions.
Results: There was no significant main effect of diagnostic group for the person-only condition. However, both computer-based versions were significantly more difficult for the ASD group: repeated measures ANOVA demonstrated significant group x condition interactions showing increased perseverative errors and fewer completed categories for the autism group.
Conclusions: We suggest that perhaps the improved performance on the computer version in the Ozonoff (1995) study arose because there was less lag time between choices, so that the ASD participants had less chance to lose the context of their previous choices in informing their next choice. In the present study, participants may have been distracted by the short videos that played in-between choices, thereby losing that context, possibly due to working memory difficulties. We discuss the results in terms of information complexity, social interactions, and neural connectivity.