Table 1.

Reconstruction Accuracy. The table summarizes the reconstruction accuracy (RA, mean across datasets ± S.E.M.) of different group ICA models. RA stands for proportion of the source variability accounted for by the ICs averaged across the 20 datasets, and was computed separately for the entire single trial images, the amplitude modulation across trials around the component peak latency, the component average timeseries, and topographies.

	Single Trial			Peak Latency			Average			Topography
Algorithm, Nr IC's	S1	S2	S3	S1	S2	S3	S1	S2	S3	S1	S2	S3
Infomax single, 20	0.98 ± .01	0.97 ± .02	0.98 ± .01	0.99 ± .01	0.99 ± .01	0.99 ± .02	1 ± .00	1 ± .00	1 ± .00	0.89 ± .03	0.89 ± .04	0.90 ± .03
Infomax, 10	0.87 ± .06	0.79 ± .08	0.66 ± .09	0.92 ± .03	0.89 ± .03	0.79 ± .05	0.99 ± .01	0.94 ± .03	0.84 ± .04	0.84 ± .04	0.82 ± .03	0.81 ± .03
Infomax, 20	0.88 ± .04	0.77 ± .07	0.54 ± .08	0.91 ± .03	0.87 ± .03	0.69 ± .06	0.99 ± .01	0.96 ± .02	0.84 ± .05	0.87 ± .04	0.83 ± .05	0.74 ± .05
Infomax, 30	0.87 ± .07	0.76 ± .07	0.50 ± .10	0.91 ± .04	0.86 ± .05	0.65 ± .05	0.99 ± .01	0.95 ± .02	0.86 ± .05	0.87 ± .04	0.81 ± .05	0.83 ± .04
Infomax, 40	0.87 ± .08	0.75 ± .09	0.47 ± .09	0.91 ± .03	0.86 ± .04	0.62 ± .08	0.99 ± .02	0.96 ± .01	0.87 ± .04	0.88 ± .03	0.79 ± .04	0.81 ± .04
Infomax, 50	0.87 ± .05	0.75 ± .06	0.44 ± .11	0.91 ± .05	0.85 ± .04	0.59 ± .07	0.99 ± .01	0.96 ± .01	0.84 ± .05	0.88 ± .04	0.81 ± .05	0.81 ± .04
ERICA, 20	0.88 ± .03	0.73 ± .05	0.46 ± .09	0.92 ± .05	0.85 ± .05	0.62 ± .08	0.99 ± .01	0.89 ± .04	0.77 ± .06	0.86 ± .03	0.76 ± .06	0.71 ± .06
JADE, 20	0.79 ± .06	0.76 ± .07	0.43 ± .10	0.85 ± .03	0.87 ± .05	0.59 ± .09	0.99 ± .02	0.96 ± .02	0.79 ± .06	0.87 ± .05	0.78 ± .08	0.77 ± .07
fastICA, 20	0.86 ± .09	0.77 ± .08	0.53 ± .09	0.90 ± .06	0.87 ± .07	0.68 ± .10	0.99 ± .01	0.96 ± .01	0.82 ± .05	0.87 ± .04	0.84 ± .06	0.74 ± .07
SIMBEC, 20	0.87 ± .08	0.78 ± .08	0.50 ± .09	0.91 ± .06	0.88 ± .06	0.65 ± .07	0.99 ± .01	0.92 ± .03	0.78 ± .06	0.87 ± .03	0.79 ± .04	0.75 ± .05