Ward's signal detection theory-based framework elucidates some aspects of interindividual differences in sensitivity, but, we argue, obscures others. Specifically, it disregards the important challenge of inferring the meaning of sensory inputs. Within Bayesian predictive coding accounts, the meaning is given by inferences to more deeply hidden causes of sensory inputs and is generally the basis for initiating context-appropriate (e.g., social) behavior. As such, when inference of hierarchical causes is hampered, as accounts of autism based on deficient precision estimation imply, a form of hyporesponsivity can emerge (together with the hypersensitivity already highlighted by Ward).
Keywords: Autism; hierarchical bayesian inference; hypersensitivity; predictive coding.