The effect of search time on perception

A group of 100 carefully selected chest radiographs was read by ten observers, five experienced and five inexperienced. The radiographs were chosen to present the readers with a disproportionately large number of both subtle abnormalities and nonpulmonary lesions. Each reader was allowed to search the radiographs for as long as appropriate, up to a maximum of four minutes. The length of time taken for each observation was recorded to the nearest second. The time-perception data were plotted on both linear and semilogarithmic graphs. The results showed that experienced readers concluded their visual search while they were still making a significant number of true-positive observations and while the true-positive detection rate was higher than the rate for false-positives. For lesions in the central portions of the radiograph (heart, lungs, and pleura), the time-perception curves were biphasic, with both a rapid and a slow component of perception. If these data are plotted on a semilogarithmic scale, each of the two components plots as a straight line. For lesions in the periphery of the radiograph (chest wall and upper abdomen), the time-perception curve is monophasic, showing only a slow component.