Fig. 3: Decoding luminance polarity and hue from MEG data: results.

a Classification accuracy as a function of time after stimulus onset for the luminance-polarity problems (center of error shading shows the average of identity problems, solid line; average of generalization problems, dashed line). Traces were generated by averaging 1000 bootstrapped samples across 18 participants. Shading shows the SE of the bootstrap samples. The stimulus duration was 116 ms (gray bar). Inset shows the difference in peak for the 1000 bootstrapped comparisons: identity minus generalization (mean = −8 ms, p = 0.34; one-sided test of the proportion of these differences that was less than or equal to zero, not corrected for multiple comparisons). Initial peak of the identity problems, solid vertical gray line, 104.4 ms [100, 110]; initial peak of the generalization problems, dashed gray line, 96.5 ms [85, 100]. Open arrowhead shows the second decoding peak, which corresponds to stimulus cessation. The horizontal sequence of data points above the x-axis, demarcated by asterisks, show time points at which decoding was above chance (determined by a permutation test across subjects and cluster corrected, see “Methods” section); onset of significant decoding for the identity problems was achieved at 65 ms [60, 80], and for the generalization problems at 55 ms [50, 70]. b The average performance across 12 sets of identity problems (solid line) and 12 sets of generalization problems (dashed line). The inset shows the difference in peak (for the 1000 bootstrapped comparisons across participants, identity minus generalization, mean = −3 ms, p = 0.82, proportion of differences that was less than or equal to zero, not corrected for multiple comparisons). The time to peak was the same for the identity and generalization problems (identity problems: solid vertical line, 121.8 ms [120, 130]; generalization problems: dashed vertical line, 119.1 ms [115, 130]). Onset of significant decoding for the identity problems was achieved at 80 ms [75, 85], and for the generalization problems, at 115 [35, 355]. Other conventions as for panel b.