Abstract
Several predictive coding theories of autism suggest that the difficulties observed in the autistic phenotype are a product of an imbalance of precision allocation between bottom-up and top-down signals. One proposed mechanism for precision allocation is attention. In two online experiments, we assessed the effects of independently modulating attention and expectation in a modified Posner paradigm using biological (BM) or coherent motion (CM) as the attentional cues. The expectation was modulated block-wise by explicitly instructing participants that 75% of targets would appear on one side of the screen, or that there was no prediction of target location. Thus, expected targets appeared on the side congruent with the block-wise cue, and unexpected targets on the opposite. Performance on the task was correlated with scores on the short Autism Quotient (AQ) questionnaire to investigate how participants’ autistic traits affect their performance. Preliminary results show that in the BM task (N=40), attended and expected targets were detected faster compared to attended but unexpected targets or when no expectation was set. The difference between attended and unattended targets was larger in both the expected and unexpected conditions in comparison to when no expectation was set. In the CM experiment (N=37), we observed inhibition of return where participants were slower at detecting attended targets as opposed to unattended. This was true for all expectation conditions. Expectation still showed a main effect, leading to faster detection of expected compared to unexpected targets. At this point, the effects of AQ were only observed in the BM experiment, where higher AQ scores led to slower reaction times. Although we cannot draw a conclusion about the effect of AQ on performance at this stage, these preliminary results suggest that expectation primes responses, above the effect of attention, and hinders performance if information is not expected despite being attended.