Stimuli and design in sequence blocks and pattern estimator blocks. (a) Illustration of sequence blocks containing full and partial sequence trials. In full sequence trials the white dot was presented for 100 ms at one of four positions successively in a fixed order with an inter-stimulus interval (ISI) of 200 ms. During partial sequence trials only one dot of the sequence (equal probability for each position in the sequence) was presented for 100 ms and all intervals remained the same as full sequence trials. (b) Illustration of target (i.e., gray dot detection) frame. (c) Graphical illustration of the eight possible dot locations, out of which only four were shown in full and partial sequence trials. (d) During random pattern estimator blocks, the white dot was presented at one of eight possible locations (as illustrated in Figure 1c) for 100 ms in random order with an ISI of 600 ms in long-ISI blocks and of 100 ms in short-ISI blocks.

Pattern estimator position (four dots from the sequence) decoding results. The decoder was trained and tested at the same time point with posterior electrodes (a), or gaze positions (b) as features. Horizontal dashed lines indicate chance level while vertical dashed lines indicate dot onset. Green and red solid lines represent long- and short-ISI pattern estimators, respectively. Shaded region represents standard error of the mean (SEM, same applied to subsequent figures). Corresponding color bars below the x-axis indicate clusters that significantly differed from chance while the black bar indicates clusters with a significant difference between long- and short-ISI pattern estimators (p < .05). (c) Temporal generalization matrix, with posterior electrodes as features, showing classification performance as a function of training and test time point in long- and short-ISI blocks. Solid black lines indicate clusters that significantly different from chance (p < 0.05). All corresponding results that discriminated eight positions can be observed in Figure 2 – figure supplement 1.

Position decoding and ERPs results for present dots. (a) Decoding of present dot positions (i.e., BCD) based on long-ISI pattern estimator (training and testing at the same time point) using posterior electrodes. (b) ERPs for the 8 posterior electrodes contralateral to each present dot position, collapsed over dots. The pink and blue lines represent full and partial sequence trials, respectively. Corresponding color bars below the x-axis indicate clusters that significantly differed from chance while black bars indicate clusters with a significant difference between full and partial sequence trials (p < .05).

Position decoding results of partial sequence trials. Temporal generalization of decoding absent predecessor (a), present (b), and absent successor (c) positions of partial sequence trials based on long-ISI pattern estimator using posterior electrodes. Solid black lines indicate clusters that significantly different from chance (p < 0.05).

Successive position decoding results of partial sequence trials. Temporal generalization of decoding the first successor (a) and second successor (b) positions based on long-ISI pattern estimator using posterior electrodes. Solid black lines indicate clusters that significantly different from chance (p < 0.05).

Pattern estimator position (all eight possible positions) decoding results. The decoder was trained and tested at the same time point with posterior electrodes (a), or gaze positions (b) as features. Horizontal dashed lines indicate chance level while vertical dashed lines indicate dot onset. Green and red solid lines represent long- and short-ISI pattern estimators, respectively. Shaded region represents standard error of the mean (SEM). Corresponding color bars below the x-axis indicate clusters that significantly differed from chance while the black bar indicates clusters with a significant difference between long- and short-ISI pattern estimators (p < .05). (c) Temporal generalization matrix, with posterior electrodes as features, showing classification performance as a function of training and test time point in long- and short-ISI blocks. Solid black lines indicate clusters that significantly different from chance (p < 0.05).