A cued reaction time task was used to test the hypothesis that there is an 'object-based' component to shifts of attention mediating the detection of luminance increment targets. The test stimulus consisted of two intersecting triangles forming a 'Star of David'. In two experiments, the cue was a brief brightening of one triangle. The target (a bright green dot) appeared on one of the triangles after a delay of 100, 200 or 500 msec. In one experiment, the target was more likely to appear on the cued triangle. In a second experiment, there was no contingency between cue and target. In both cases, reaction times to targets which appeared outside (but not inside) the cued triangle were more than 10 msec longer than other targets, but only at the shorter cue-target delays. This indicates that the attentional system which regulates luminance increment detection cannot select the cued triangle. It appears that the attentional spotlight can be briefly deformed into a triangular shape, and that it is the rapid, fast-decaying and reflexive exogenous system, rather than the slower acting, persistent and voluntary endogenous system, that mediates this effect. A third experiment using a central, symbolic cue showed no significant cue-validity effects, indicating no contribution from the endogenous system. It is concluded that tasks requiring only stimulus detection cannot unequivocally discriminate between spatial and object-based components of attention.