Attention Switching vs Attention Sharing in Searching Dual RSVP Streams
Ideas about how visual attention is distributed over space include spotlight, zoom lens, and various resource allocation models. Spotlight and serial allocation models assume that attention is narrowly focused and switches from one object to another in visual search. Zoom lens and parallel allocation models, on the other hand, describe a flexible gradient within which attention can be shared among several items simultaneously. We report two experiments in which simultaneous rapid serial visual presentations (RSVPs) of two streams of digits were used, one above and one below a fixation point. In experiment 1, subjects were told to report the digit immediately following a uniquely coloured signal digit. In some trial blocks the coloured signal digit always appeared in either the top or bottom stream, and in other blocks the signal digit could occur in either stream. Stream location probabilities were varied between blocks in order to induce strategic variations in attentional allocation. In experiment 2, subjects were told to report the first two digits visible when the fixation point changed colour. Subjects were instructed to report one digit from the top stream and one from the bottom, with report order counterbalanced between blocks. The lag between the response signal and the actual digit reported was shown to vary strongly with signal location probability (experiment 1), and the lag between items reported from the top and bottom streams depended heavily on the order of report (experiment 2). The results were more consistent with an attention-switching model than with an attention-sharing model of visual attention.