A time-based two-stage model of visual selection: Evidence from standard visual search tasks

How does the visual system achieve efficient information selection under limited processing capacity is a fundamental question in cognitive psychology. Through three experiments, the present study reveals a time-constrained two-stage visual selection mechanism. Experiment 1's color search tasks demonstrated bilinear Search Time × Set Size functions, with a clear kink point at 50-100 ms after stimulus onset, showing shallower slopes (2.7-10.4 ms/item) before versus steeper slopes (10.7-20.7 ms/item) after, indicating two distinct processing stages. Experiment 2 extended these findings to digit search: despite capacity differences (color = 7.6 vs. digit = 3.8 items), the critical kink point time remained stable (60-70 ms), suggesting cross-stimulus generalizability. Experiment 3 disrupted this bilinear pattern by delaying search onset, confirming its time dependence. Individual analyses showed unimodal kink point time distributions (peak 60-80 ms) but multimodal set-size distributions, indicating time-driven stage transitions. These findings establish that visual selection has a time-based parallel-to-serial transition, and the early parallel stage has a stimulus-dependent capacity. This two-stage selection model integrates the core tenets of the Guided Search model and attentional window theories, offering a new perspective for the architecture of visual information processing. (PsycInfo Database Record (c) 2026 APA, all rights reserved).