Visual search in multicharacter arrays with and without gaps

Subjects searched through briefly presented arrays of letters in a controlled order, indicating quickly which of two possible targets had occurred. Some arrays contained gaps—three missing letters. Reaction time (RT) and accuracy were both improved by a gap after the target; improvement was less when the gap preceded the target. To account for these results, a new model is proposed, one which calls for overlapping processing of successive array items. This is not a “hybrid” model, but a third alternative between the two extremes of serial (zero overlap) and parallel (complete overlap). Quantified, the overlapping model generates U-shaped serial-position curves and produces RT predictions in good accord with our data from arrays with and without gaps. The predicted functions for RT vs. array size are concave upward; however, for arrays of five or less they are virtually linear and not very different in slope for positive and negative trials. Although this model is primarily designed for RT, with some additional assumptions it can be extended to accuracy results.