Duncan interpreted the fast response time as suggesting that people perceive each square as a discrete, coherent unit and record various information about it together, allowing for quick recall.

But all the research that has fleshed out his work over the years has shared one major problem: the objects that the experiments contrasted were in different locations, so the assay could not completely discriminate between the location- and object- based theories of attention.

It took a combination of the expertise on visual attention of Rutgers Professor Zenon W. Pylyshyn and experimental methodology developed by Holcombe in the Harvard Vision Sciences Laboratory to resolve this problem.

For the research that led to his Ph.D. dissertation, Holcombe developed a technique in which two Gabor patches--spinning, ridged, colorful circles made out of sine waves--are superimposed on top of each other on a computer screen.

His technique is a modernization of the classic optical illustration developed by Edgar Rubin in which viewers see either a vase or two faces depending on how they direct their attention. The viewers perceive one of the two patches as being in the foreground and the other as being in the background by choosing where to direct their attention.

At Rutgers, the vision scientists exploited Holcombe's technology in experiments where a viewer was asked to pay attention to one of the two Gabor patches as the color, spin and stripe size of both were changed randomly.

Sometimes, the randomly changing attributes overlapped--both patches turning red at the same instant for example--which made following along a little tricky. But 90 percent of the time, subjects could identify which patch was which at the end of the experiment.

Because the features of the patches changed as the experiment went on, the researchers ruled out the possibility that observers kept track of them by a constant difference in their characteristics.