Article Figures & Data
Figures
- Figure 1.
Task design. The configuration of sequence types, number of trials, transition probabilities, and sequence probabilities in four blocks. The tone icons colored in black and red represent tone x and y in different pitches, while the tone icon with the dashed outline represents omission (“o”, no tone delivered). The stimulus onset asynchrony (denoted as SOA) represents time interval between the onsets of a tone and the next. The inter-trial interval (ITI) represents time interval between the offset of one sequence's last tone and the onset of the next sequence's first tone. The probabilities were rounded to two decimal places.
- Figure 2.
ERPs and deviant responses from contrasts. A, The group mean ERPs of sequences xxx, xxy, and xxo in Block 1 at channel Cz. Solid vertical lines represent onsets of the stimuli of a sequence. Time zero was set to be the onset of the last stimulus. B, Contrast responses obtained by contrasting ERPs of sequences xxy and xxx in 4 blocks. The gray shade represents the time range of the MMN. C and D, Peak amplitudes and latencies of the MMN at Cz in 4 blocks. The block order was sorted from low to high probability in terms of TP(y|x) and SP(xxy). The 30 dots in each block correspond to participants. The box plot represents the median (red horizontal line), quartiles (the bottom and top edges), 95% confidence interval (notches), and the maximum and minimum (black horizontal lines). Results from repeated measures ANOVA and Spearman's rank correlation analyses are shown above the plots. The red line represents a significant difference between two blocks (*<0.05; ***<0.001).
- Figure 3.
Neural signatures of prediction-error signals. We resampled 100 datasets, each containing contrast responses of 4 blocks, and then performed PARAFAC decomposition for each dataset. For the two extracted subcomponents, activations are shown in (A) Channel, (B) Timecourse, and (C) Block dimensions. The Block dimension represents activation of the subcomponents to the responses (the blue line). The error bar represents the standard deviation. The model values of the local and global prediction errors are denoted as PE1 and PE2 (the green line).
- Figure 4.
Reconstruction with functional assignments and their differences from the actual ERP. A, Examples of reconstructed signals at Cz in Block 1. The black line represents the actual contrast response. The blue line represents the reconstructed response with Subcomponent 1 as PE1 and Subcomponent 2 as PE2. The orange line represents the reconstructed response with Subcomponent 1 as PE2 and Subcomponent 2 as PE1. The yellow line represents the reconstructed response with Subcomponent 1 and 2 both as PE1. The purple line represents the reconstructed response with Subcomponent 1 and 2 both as PE2. Mean squared differences in panel B and Pearson correlation coefficients in panel C were calculated then averaged across channels and blocks. The 100 dots for each assignment represent resamples. Results from repeated measures ANOVA and pairwise comparisons are shown above the plots.
Extended Data
Figure 2-1
Download Figure 2-1, TIF file.
Figure 2-2
Download Figure 2-2, TIF file.
Figure 2-3
Download Figure 2-3, DOCX file.
Figure 3-1
Download Figure 3-1, TIF file.
Figure 3-2
Download Figure 3-2, TIF file.
In this issue
