Deviant sounds yield distraction irrespective of the sounds' informational value

Oddball studies show that rare and unexpected changes in an otherwise repetitive or structured sequence of task-irrelevant sounds (deviant sounds among standard sounds) ineluctably break through attentional filters and yield longer response times in an ongoing task. Although this deviance distraction effect has generally been thought of as an involuntary and adaptive phenomenon, recent studies questioned this view by reporting that deviance distraction is observed when sounds predict the occurrence of a target stimulus (informative sounds) but that it disappears when sounds do not convey this information (uninformative sounds). Here, I challenge this conclusion and suggest that the apparent absence of deviance distraction with uninformative sounds results in fact from 2 opposite effects: deviance distraction when the previous trial involved a target and required responding, and a speeding up of responses by deviant sound following trials involving no target and requiring the withholding of responses. Data from a new experiment, new analyses of the data from 3 earlier studies, and the modeling of these data all converge in suggesting the existence of deviance distraction impervious to the sounds' informational value. These results undermine the proposition of a late top-down control mechanism gating behavioral distraction as a function of the sounds' informative value. (PsycINFO Database Record