Electronic Supplementary Material

Supplementary Table 1. Individual grouping sensitivity calculated using two methods.

Grouping sensitivity calculated using responses a, b, c, and d (as for Figure 3) / Grouping sensitivity calculated using responses a and b alone
SG / 11.12 / 11.12
JH / 4.67 / 5.26
MK / 6.01 / 6.28
YY / 10.09 / 10.37
SS / 4.96 / 5.11
HK / 2.78 / 3.88
IM / 13.23 / 23.71
NK / 9.55 / 9.83
MH / 4.83 / 5.63
OM / 11.00 / 13.02
TK / 2.39 / 4.84
MC / 9.15 / 9.38
KH / 2.19 / 3.33


Topographical analysis of C1 and P1 activity across aspect ratios

We reported the results of topographic analysis in section “Peak C1” of “Results” for a single condition (aspect ratio 1.0 and orientation 157.5°), for which the amplitude of peak C1 was maximal. Using the condition for which the electrical activity is most pronounced allows one to observe topographic differences most clearly. We also asked whether we can obtain additional information about topographic distribution of activity from other experimental conditions, at which C1 activity was less pronounced. We did the same topographical analysis in 4x7 array of electrodes as described in the main text, but now we included aspect ratio as an additional factor in the ANOVA (the four orientations were averaged). To make the amplitudes compatible across AR conditions, we normalized the amplitude in each condition so that the RMS values of across-subject average amplitudes from different conditions were the same, as proposed by McCarthy and Wood (1985). The 4x4x7 repeated-measures ANOVA showed the only significant effect of aspect ratio (F(3, 15) = 5.0, p < 0.05, ε = 0.92), while the sagittal and frontal topographical factors and their interactions were not significant. Fig. S1A illustrates that the reported inter-hemispheric difference of C1 (Fig. 3B) is observed only for the condition in which C1 amplitude was most pronounced and which we selected in the main analysis (AR=1.0). This indicates that low-amplitude conditions do not offer additional information about topography of C1 activity, presumably because of the low signal-to-noise ratio.

In contrast, signal-to-noise ratio for P1 component is much higher than for C1, so that the same topographic pattern is observed in all AR conditions (Fig. S1B). An ANOVA on the P1 amplitude (normalized across all AR conditions as above for the C1 analysis) revealed a significant interaction between the sagittal and frontal directions of the array (F(18, 216) = 3.2, p < 0.01, ε = 0.43), very similar to what we found in the analysis reported in the main text (Fig. 3C).

This analysis indicates that it is advisable to do topographic analysis of ERP activity using conditions in which the activity is most pronounced. For analysis of low-amplitude activity such as C1 this selection is critical, which is why we use this approach for both low-amplitude and high-amplitude activity.

INSERT FIGURE S1 HERE

Figure S1. A. Mean-error plots of the normalized amplitude of C1 activity for four aspect ratios (averaged across orientations) in the high-sensitivity group. B. Mean-error plots of the normalized amplitude of P1 activity for four aspect ratios (averaged across orientations) for all participants. In both panels, the amplitudes are shown for 28 channels organized into 4 chains of 7 electrodes (colored green in Fig. 3D). Numbers on the abscissa correspond to the ordinal numbers of electrodes within each chain, from left to right in the frontal plane.

McCarthy G, Wood CC (1985) Scalp distributions of event-related potentials: an ambiguity associated with analysis of variance models. Electroencephalogr Clin Neurophysiol 62: 203-208

3/3