Fig. 2: Behavioral and decoding results (N = 24).

In box plots: grey lines indicate individual participant results; shaded box extends over IQR; the middle line represents mean; whiskers extend to mini/maximum values; diamonds indicate condition means; notches indicate confidence interval of median; plot size indicates data volume; all stars represent results of two-sided, preregistered t-tests; * = p < 0.05; ** = p < 0.01; *** = p < 0.001. a Participants were faster to respond to targets at high-probability (HP) locations than at low-probability (LP) locations. b Participants were faster when targets were presented at the same location on sequential trials (repeat trials) than when they switched to a new location (switch target). c Participants were slower when distractors were present than when they were absent. Participants were, additionally, especially slow when distractors were present at the HP target location. d Decoder results comparing ping and no-ping trials. Data was baselined in the −200 to 0 ms window pre-ping onset. Shaded areas represent the standard error of participant means. Lines are smoothed using the scipy function gaussian_filter with an alpha of 1.5. Lower red bars represent significant clusters identified where decoding of ping trials was above chance. Black bars represent clusters identified in which ping and no-ping decoding differed significantly. Posthoc analysis additionally showed that all four locations contributed to this above chance decoding (see Supplementary Fig. 5 for the confusion matrix, as well as decoding without boosting as per preregistration).