ABSTRACT
The reaction time for shifting attention from a target letter at the left of fixation to a stream of numerals at the right of fixation is measured by noting the earliest-occurring numeral an observer can report. In addition to this attention reaction time (ART), the observer produces a conventional motor reaction time (MRT) by indicating target detection with a finger movement. ARTs and MRTs have comparable distributions over trials, but ARTs vary more with the difficulty of targets than do MRTs to the same targets. The results are accounted for by a two-stage three-component model of reaction times consisting of a shared detection component (in which detection for attention responses requires 50% longer processing than detection for motor responses) followed by independent attention and motor response-generating components. A metal snapshot model is proposed to account for the functioning of the attention response-component.
