Abstract
The classic work of Jonides & Yantis (1988 Perception & Psychophysics, 43, 346–354) showed that attention was not captured by elements having a unique color, shape or luminance. Subsequent work by Theeuwes (1992 Perception & Psychophysics, 51, 599–606) demonstrated that in some conditions static singleton do capture attention and that this capture is not under top-down control. One possible explanatory mechanism is to assume that observers are able to either focus their attention on a limited area in visual space or to spread attention across the visual field. In the current study we show that in a task in which observers have to focus attention, salient singleton outside the focus of attention do not capture attention anymore. However, when this very same task is used and observers are instructed to spread their attention, static singleton capture attention. Our findings have implications for models of visual attention. We propose that salience computations across the visual field are more or less restricted to the attentional window of the observer. The size of the attentional window appears to be under top-down control. However, within the attentional window salience computations occur in an automatic, bottom-up fashion, and the location having the highest salience is prioritized (cf. Itti & Koch, 2000 Vision Research, 40, 489–1506).