Paradigm overview.

A. A single trial, with two example images. Images were presented for 100 ms each with 800 ms interstimulus interval, and an intertrial interval of 1300-2200 ms. Participants were required to respond to upside down images with button press; all upside down images were trailing. B. Transitional probability matrix determining category pairs. Five leading categories and four trailing categories were used. Orange represents the 2:1 condition, and green represents the 1:2 (control) condition. Cells with dots represent the valid pairs, cells with Xs represent the invalid pairs, and empty cells represent non-existent pairs. C. Reaction times across Valid, Invalid, and Neutral conditions. Asterisks indicate significant results at p<0.05 after correction for multiple comparisons.

Results of data analysis.

A. Grand average ERP statistical analyses in a cluster of occipital channels at Channel Oz. Significant difference between valid and invalid trial amplitudes. B. Topography plot of the t-values resulting from paired t-tests comparing Valid and Invalid EEG data at 150 ms. C. SVM decoding results for Leading image sensory decoding. Dashed vertical line at 0 s indicates stimulus onset. D. Decoding results for Trailing image sensory decoding. Dashed rectangles denote latencies of significant effects (pFWE < 0.05). Dashed vertical line at 0 s indicates stimulus onset. E. Decoding results for Trailing image memory decoding. Dashed rectangle denotes latencies of significant effects (pFWE < 0.05). Dashed vertical line at 0 s indicates stimulus onset.

Analysis of validity effects on decoding over trial bins.

Error bars indicate SEM. Dashed horizontal lines indicate chance level. Asterisks indicate significant results at p<0.05 after correction for multiple comparisons. A. Learning over time at 123-180 ms in sensory decoding. B. Learning over time at 280-196 ms in sensory decoding. C. Learning over time at 280-296 ms in memory decoding.