Abstract
Visual attention can select objects as well as locations, and this object-based selection operates on both grouped-array and space-invariant levels (Vecera & Farah, 1994). Space-invariant selection has been typically examined by the feature-report task: Observers are presented with a box and a line that are superimposed. Each object has two attributes (height and gap for a box and tilt and texture for a line). Observers' accuracy is higher when two attributes are part of one object rather than two objects. This finding closely matches the properties of visual short-term memory (VSTM): Visual objects were remembered based on a number of objects rather than a number of object's features (Luck & Vogel, 1997). We demonstrated that space-invariant selection and VSTM shared the same underling mechanisms by loading the feature report task with a concurrent VSTM task (OPAM, 2005). Different concurrent VSTM tasks led to different interference patterns. Under an object load, observers exhibited an increased object-based effect for the second response trials. However, when the information of 2 to-be-reported attributes was withheld until after the object was masked, the object-based effect was abolished, although small object-based effects returned when salient object cues were available. We hypothesized that, when there was no salient object cue to indicate which object each attribute belonged, these floating attributes slowed the integration of features into an object file. In the current study, we examined whether the number of attributes makes a difference for the presence of space-invariant object-based selection.
Sigma-Xi Grant-In-Aid of Research to Michi Matsukura, NIH R01 MH63001 to Steve Luck, and NSF BCS 03-39171 to Shaun Vecera.