Fig. 1: Trial timecourse and experiment design.

a Example of a stimulus display. Note that the yellow markings were not present in the actual experiment. Participants were tasked to find a uniquely shaped shape singleton and report the embedded line’s orientation using the ‘z’ key for horizontal or ‘/‘ key for vertical. Colors and shapes randomly varied between trials, keeping the target characteristics unknown. Salient color singletons served as distractors in 2/3 of the trials. b Experiment block order and decoding regime. Biased blocks had one significantly more likely target location (37.5% of trials), while neutral blocks had no high-probability location, distributing targets equally among eight locations. Participants encountered four biased blocks followed by a neutral block. The subsequent four biased blocks used different high-probability locations, cycling among four (top, bottom, left, and right) with the order counterbalanced across participants (the order shown in the figure is just one example). The rightmost panel presents the ‘dummy decoding’ regime: as a control analysis, decoders received fake category labels representing random time intervals without a consistent high-probability location. Temporal features of EEG signals were preserved while removing categorical features of a shared high-probability location. c Examples of the two types of pings used in the experiment. d Timecourses for no-ping (top) and ping (bottom) trials. Pre-stimulus periods were time-matched (1500–2200 ms). Trials began with a black screen, followed by a fixation dot. In half the trials, a salient ping appeared for 200 ms. The stimulus display required participants to identify a unique shape and report the line orientation, lasting until a response or 2500 ms. Trials without pings still recorded a trigger event for baselining in the no-ping decoding analysis.