Abstract
Distinctive aspects of a visual display can grab our attention. This “attention capture” is particularly strong for new objects. Whether new objects per se, or the transient signals that often accompany them provide the basis for capture has been debated, however. We examined whether new objects capture attention in natural scenes, and focused on the importance of a transient signal. While viewing a scene, a new object appeared during a fixation such that it was a transient, or during a saccade, which because of saccadic suppression, eliminated the transient signal. Capture was measured by the eyes' propensity to be directed to the new object. In the fixation condition, 63% of the eye movements immediately following the onset were directed to the new object which accounted for 70% of first looks to the new object. The fixation during which the onset occurred was, on average, 95 ms shorter than all other fixations and the first saccade to the new object subtended 4.5 degrees more than the average of all other saccades indicating the onset was interruptive and “eye grabbing.” In the saccade condition, 18% of the first saccades initiated after the onset were directed to the new object which accounted for 33% of first looks to the new object. In a control condition, where no onsets occurred, 10.4% of fixations were on the same critical object. The fixation just after the onset was longer than other fixations suggesting that a change to the scene was implicitly detected, although not necessarily localized. Indeed, the majority of fixations on the new object occurred after several saccades indicating detection of the new object is guided by VSTM. Thus, new objects can receive attentional prioritization during scene viewing via different mechanisms: Transient objects capture attention quickly and reliability, whereas without a transient signal, new objects are prioritized over several saccades as VSTM works to explicitly identify the change.
Funding by the National Science Foundation (BCS 0094433 & IGERT DGE 0114378) and the Army Research Office (DAAD19-00-1-0519).