The time courses of attention switching between and within auditory and visual modalities were measured with a sensitive attention reaction time paradigm based on rapid serial presentation (1). In each trial, one digit stream and one letter stream were presented at 6 items/s. Observers identified a cue stimulus (2 or 9) embedded in a digit stream, switched to a letter stream, and then reported the following 4 letters and the identity of the cue. In intra-modal visual to visual (VV) and auditory to auditory (AA) conditions, the streams were presented at different foveal locations or in different ears. In cross-modal visual to auditory (VA) or auditory to visual (AV) conditions, the cue was in one modality and the letters in the other. Payoffs encouraged the report of early letters following the digit cue. The time course of switching is measured by the probabilities of reporting items in particular locations. For the cross-modal conditions, the report began with items simultaneous with the cue. In the intra-modal conditions, the report began with items immediately following the cue, but rarely included the simultaneous item. This time-course advantage for cross-modal relative to intra-modal switching is reflected in the medians of the temporal location of the items reported: 91 ms for VA and 125 ms for AV; 320 ms for VV and 305 ms for AA. We conclude that attention switching between two sensory modalities is approximately 200 ms faster than attention switching within a single sensory modality. A reanalysis of the data in terms of stimulus position relative to the first-reported item, effectively compensating for the overall temporal shift of the whole response ensemble, reveals that although cross-modal cuing is faster than intra-modal cuing, the full report distributions can be accounted for by a single attention gate which opens with different delay distributions.

(1) Reeves & Sperling, Psych Rev, 1986.